1General framework: the sensory epithelium as the substrate of transduction; scala media, basilar membrane and geometry of the cochlear duct

Among mammalian sense organs, the cochlea is the most refined example of mechanoelectrical transduction organised on a strict topographic principle, in which the gross geometry of the ductus cochlearis deterministically sets the biophysical properties of the sensory epithelium it contains ^[5,27]. Understanding the microscopic anatomy of the inner ear means, first of all, understanding how three concentric fluid compartments, the scala vestibuli, the scala media (or scala media cochleae) and the scala tympani, form a hydromechanical system in which acoustic energy transmitted by the stapes is converted into a travelling wave along the basilar membrane, and finally into a bioelectrical signal at the hair cells ^[27,26]. This introductory chapter lays the conceptual and morphological foundations for the later sections on the hair cells, the supporting cells, the stria vascularis and the ganglion neurons.

1.1 The cochlear duct and its spiral geometry

The cochlear duct develops as a tubular structure coiled around the modiolus for about two and three quarter turns in humans, with a progressive decrease in the width of the basilar membrane from base to apex, a decrease that is paradoxically accompanied by an increase in its mechanical compliance ^[27,5]. This inverse relation between size and stiffness is the structural correlate of cochlear tonotopy: high frequencies reach their maximum amplitude of vibration in the stiff, narrow basal region, while low frequencies dissipate progressively towards the wide, compliant apex ^[27,26,3]. The scala media, lying between Reissner’s (vestibular) membrane above and the basilar membrane below and laterally, is the endolymphatic compartment proper, an electrochemical environment unique in the body, characterised by a high potassium concentration and a strongly positive endocochlear potential, generated and maintained by the stria vascularis ^[28,29,30].

1.2 The basilar membrane as the mechanical substrate of tonotopy

The basilar membrane is not a homogeneous structure but a composite of radially arranged collagen fibres embedded in a proteoglycan matrix, whose stiffness varies longitudinally by about three orders of magnitude between base and apex ^[27,24]. On it rests the organ of Corti (organum spirale), the neurosensory epithelium proper, whose cellular architecture, with inner and outer hair cells, inner and outer pillars, Deiters cells and Hensen cells, is organised so as to transform the passive wave motion of the cochlear partition into an amplified and frequency-tuned signal ^[3,5,9]. The travelling wave described by von Békésy, later revised in the light of the discovery of the cochlear amplifier, does not by itself explain the fineness of human auditory tuning, of the order of fractions of a semitone: an active process must be invoked, dependent on the integrity of the outer hair cells and their prestin-mediated somatic motility, which feeds back on basilar membrane motion, amplifying it selectively near its local characteristic frequency ^[18,20,23].

1.3 The sensory epithelium as an integrated biomechanical structure

From an anatomical and functional perspective it is essential to regard the organ of Corti not as a simple carpet of sensory receptors next to passive supporting cells, but as an integrated biomechanical tissue in which the supporting cells, particularly the pillar and Deiters cells, contribute actively to stiffness and to force transmission along the cochlear partition ^[9,10,12]. The microtubular cytoskeleton of these cells, whose organisation depends on regulators such as the GAS2 protein, forms a structural scaffold without which the transmission of sound energy across the cochlear partition is impaired, with consequent hearing loss even without direct damage to the hair cells ^[10]. This integrated view is further supported by the observation that selective ablation of pillar and Deiters cells in the mouse severely disrupts postnatal development of the cochlea and of hearing, even when the hair cells initially survive intact ^[9,11].

1.4 The scala media and the ionic environment of transduction

Physiologically the scala media is a compartment of its own kind: although it is an extracellular space, its ionic composition resembles that of the cytosol, with potassium concentrations close to 150 mmol/l and a low sodium concentration ^[28,30]. This peculiar composition, together with the endocochlear potential of about +80 mV generated by the stria vascularis, creates a transepithelial electrochemical gradient that drives potassium into the hair cells through the mechanoelectrical transduction channels at the tips of the stereocilia ^[28,29,26]. Loss of this potential, as occurs in strial dysfunction related to ageing or ototoxic insults, drastically reduces hearing sensitivity regardless of the state of the hair cells themselves, underlining that sensory function depends on a multicomponent tissue architecture and not on the integrity of the receptor alone ^[1,31,33].

1.5 Ontogenetic perspective and clinical implications of the geometric model

The geometry of the cochlear duct is not static over time: during fetal and neonatal development, maturation of the endocochlear potential proceeds in step with the structural differentiation of the stria vascularis and the organ of Corti, a process whose timing differs substantially between altricial and precocial species but which in humans is largely complete before birth ^[34]. Understanding this fundamental architecture, scala media, basilar membrane and spiral geometry of the duct, is indispensable for correctly interpreting the pathological pictures discussed in later chapters, from presbycusis related to strial dysfunction ^[1,31] to platinum ototoxicity with its tropism for the stria vascularis and the spiral ganglion ^[2], to the genetic syndromes that selectively affect the ion channels responsible for endolymphatic homeostasis ^[33]. In this sense the microscopic anatomy of the inner ear cannot be separated from its physiology: every structure, from the sensory epithelium to the strial vascular network, from the basilar membrane to the spiral ganglion, finds its raison d’être in a unitary functional design aimed at the faithful, amplified transduction of the acoustic signal across the whole spectrum of frequencies audible to humans.

2The organ of Corti: overall architecture

In its histological compactness the organ of Corti is one of the most finely organised sensory epithelia in the human body, a cellular mosaic in which the topographic arrangement of each element responds to precise biomechanical requirements for the transduction of the sound wave into a neural signal^[5]. The structure rests on the basilar membrane, whose gradients of width and stiffness run in opposite directions along the base-to-apex axis: narrow and stiff at the base, where high frequencies resonate, wide and compliant at the apex, where low frequencies are coded^[26][27]. This mechanical gradient is not a mere descriptive anatomical fact but the physical substrate of cochlear tonotopy, whose functional expression depends critically on the integrity of the overlying active amplification apparatus^[24]. On the basilar membrane stands, like an architectural scaffold of support and connection, the assembly of sensory and supporting cells, topped by the reticular lamina, a cohesive apical surface that seals the endolymphatic space, separating it sharply from the underlying perilymphatic compartment^[3].

Fluid compartments and intraepithelial spaces

Within this architecture there are cavities of considerable functional importance: the tunnel of Corti, bounded by the inner and outer pillars, and the spaces of Nuel, which surround the outer hair cells and give them the mechanical freedom indispensable to their motility^[3][9]. These spaces, filled with a fluid whose ionic composition resembles perilymph, allow the vibrations generated by the somatic motor of the outer hair cells to propagate freely without the constraint of a dense cellular environment^[17]. Laterally, the inner and outer spiral sulci complete the topographic picture, acting as a transition zone towards the non-sensory epithelium of the spiral limbus and towards the Claudius and Böttcher cells, discussed below^[5].

2.1 Inner hair cells

The inner hair cells (IHCs), arranged in a single row along the whole length of the human cochlea, are estimated at about 3,500, a seemingly small number compared with the amount of auditory information they carry^[5][26]. Their morphology, classically described as flask-shaped, is characterised by a swollen cell body containing a large central nucleus, topped by a narrow apical pole that supports the cuticle and the hair bundle, the latter arranged in rows forming an open U or V^[3]. The cuticle, a thickening of the actin-rich apical cytoskeleton, provides the mechanical rigidity needed to anchor the stereocilia and faithfully transmit the deflections produced by the sound wave^[25]. What makes the IHCs the primary sensory element of hearing is their afferent innervation: about 95% of the fibres of the cochlear nerve go exclusively to these cells, each of which receives multiple synaptic contacts from type I fibres of the spiral ganglion neurons^[6][41]. The glutamatergic synapse between the IHC and the afferent dendrite is a privileged model for understanding peripheral synaptic mechanisms, being characterised by fast, precisely timed vesicle release mediated by otoferlin and Cav1.3 calcium channels^[6][37]. Disruption of this synaptic apparatus, as shown in mouse models carrying Otof mutations, reproduces phenotypes resembling human auditory neuropathy, with preserved otoacoustic emissions but an abolished brainstem response^[37].

2.2 Outer hair cells

The outer hair cells (OHCs), more numerous, are estimated at about 12,000 and are arranged in three rows in the basal region, sometimes increasing to four or five towards the apex, a non-uniform distribution that reflects regional differences in function^[5][3]. Their elongated cylindrical shape, quite distinct from the flask-shaped profile of the IHCs, goes with a double anchorage: at the base by the phalanges of the Deiters cells, at the apex by insertion into the reticular lamina, an arrangement that allows them to oscillate longitudinally without losing the cohesion of the sensory epithelium^[7][9]. It is precisely this freedom of movement, made possible by the spaces of Nuel, that allows expression of the somatic motor, supported by the membrane protein prestin (SLC26A5), whose voltage-dependent conformational cycle produces the cell contraction and elongation underlying so-called electromotility^[18][19][17]. This mechanism, operationally defined as the cochlear amplifier, feeds back on basilar membrane vibration, increasing its sensitivity and frequency selectivity by more than 40 dB compared with a purely passive system^[22][20][21]. Selective loss of OHCs, as occurs in many forms of presbycusis and in cisplatin ototoxicity, translates clinically into a raised threshold and a loss of frequency discrimination disproportionate to the number of cells actually damaged^[1][2].

2.3 Supporting cells

Around and beneath the sensory cells lies a heterogeneous population of supporting cells, whose function goes beyond a simple architectural role to take on mechanical, metabolic and homeostatic importance of the first order^[15][10]. The inner and outer pillars, stiff because of their content of microtubules and intermediate filaments, bound the tunnel of Corti and transmit the mechanical forces generated by the travelling wave, contributing to the overall stiffness of the epithelium^[9]. The Deiters cells, first described by Otto Deiters in the second half of the nineteenth century, support the OHCs with their apical phalanges and modulate their local stiffness, taking an active part in transmitting mechanical forces to the reticular lamina^[7][12]. Recent optogenetic studies have shown that Deiters cells are not merely passive elements but contribute actively to the regulation of cochlear electromechanical responses, and can modulate the sensitivity of the organ in response to local signals^[12][13]. Laterally lie the Hensen cells, columnar and lipid-rich, followed by the Claudius and Böttcher cells, the latter involved in reabsorption and ion regulation in the outer spiral sulcus^[5][15]. The whole supporting system is held together by tight junctions which, sealing the lateral membranes of adjacent cells at the level of the reticular lamina, form the physiological barrier essential to separating the potassium-rich endolymphatic compartment from the perilymphatic and intracellular compartments of opposite ionic composition^[28][32]. Breakdown of this barrier, whether genetic, toxic or degenerative, dissipates the transepithelial electrochemical gradient with immediate consequences for mechanoelectrical transduction, evidence that the supporting cells are not accessory structures at all but indispensable functional partners of the sensory cells proper^[9][11][14].

3The stria vascularis and the endocochlear potential

The stria vascularis (stria vascularis) is the secretory epithelium of the lateral wall of the cochlear duct, responsible for producing endolymph and generating the endocochlear potential, bioelectrical conditions indispensable for the working of the mechanoelectrical transducer of the hair cells^[5,28,29]. Unlike classical secretory epithelia organised as a single polarised cell layer, the stria vascularis has a three-part architecture, unique among mammalian epithelia, in which three morphologically and functionally distinct cell populations, marginal, intermediate and basal cells, cooperate to create an extracellular compartment with a very high potassium concentration and a strongly positive electrical potential^[29,30].

3.1 Cellular architecture: marginal, intermediate and basal cells

The marginal cells line the endolymphatic surface of the stria and have extensive basolateral infoldings of the plasma membrane, which greatly increase the surface available for pumps and ion channels; this ultrastructural specialisation is the morphological basis of their active secretion of potassium ions into the endolymph^[28,29]. The intermediate cells, of melanocytic derivation and originating from the neural crest, interdigitate with the basolateral processes of the marginal cells; they express the transcription factor Mitf and essential potassium channels, and their genetic impairment in mouse and human models produces congenital deafness associated with pigmentation defects, evidence of the determining rather than accessory role of this population in strial physiology^[35,29]. The basal cells, flattened and joined by tight junctions, form a barrier that functionally separates the intrastrial compartment, a narrow extracellular space between the intermediate and marginal cells, from the connective tissue of the underlying spiral ligament^[29,30]. This intrastrial compartment, bounded by gap junctions between basal and intermediate cells, is where the characteristic electropositive potential is generated, electrically isolated from both endolymph and perilymph by high-resistance epithelial barriers^[29,31].

3.2 Transporters and ion channels: KCNQ1/KCNE1, Na⁺/K⁺-ATPase, Kir4.1

The net flow of potassium from the intrastrial compartment into the endolymph across the apical membrane of the marginal cells is mediated by the heterotetrameric KCNQ1/KCNE1 channel, whose coordinated expression is indispensable for the final secretion of K⁺; mutations of either subunit cause syndromic forms of deafness with QT prolongation, confirming the close functional dependence between the two genes^[29,30,28]. On the basolateral side of the marginal cells, the Na⁺/K⁺-ATPase and the Na⁺-K⁺-2Cl⁻ cotransporter (NKCC1) take up potassium from the intrastrial space, completing a transcellular transport circuit that concentrates potassium against an unfavourable electrochemical gradient^[30,32]. In the intermediate cells, the inwardly rectifying channel Kir4.1 (encoded by KCNJ10) mediates the exit of potassium from the cytoplasm into the intrastrial space and is regarded as the main generator of the electrical component of the endocochlear potential: its deletion or dysfunction, seen in EAST/SeSAME syndromes, almost completely abolishes the positive potential while leaving the chemical component of the ionic gradient relatively preserved^[33,29]. This experimental dissociation between the chemical and electrical components of the endocochlear potential has made it possible to attribute to Kir4.1, rather than to the K⁺ concentration alone, the role of primary determinant of the +80 mV positivity measurable in the scala media^[29,33].

3.3 Gap junctions and potassium recycling: GJB2/GJB6 and the fibrocytes of the spiral ligament

Long-term maintenance of cochlear potassium homeostasis requires not only strial secretion but also a potassium recycling system that recovers the ion after it leaves the hair cells during mechanoelectrical transduction^[32,5]. Potassium released basally by the outer hair cells into the spaces of Nuel is taken up by the supporting cells of the organ of Corti and transferred, through a network of gap junctions made of the connexins GJB2 (connexin 26) and GJB6 (connexin 30), to the fibrocytes of the spiral ligament, and finally returned to the stria vascularis through a second gap junction network between fibrocytes and basal and intermediate cells^[32,30]. Mutations of GJB2 are the most frequent monogenic cause of non-syndromic sensorineural hearing loss, which underlines the clinical relevance of this extracellular circuit, often overlooked in favour of the stria proper^[32]. The spiral ligament, traditionally divided into five types of fibrocytes (I to V) on topographic and immunohistochemical grounds, is not simply a supporting connective tissue but takes an active part in ion transport: type II and V fibrocytes, close to the stria, express high densities of Na⁺/K⁺-ATPase and Na⁺-K⁺-2Cl⁻ transporters, while the more basal type I fibrocytes are electrically coupled to them through gap junctions and help transfer recycled potassium to the intrastrial compartment^[30,32].

3.4 Generation of the positive endocochlear potential

The integrated result of these transport mechanisms is the generation in the scala media of an electrical potential of about +80 mV relative to the perilymph, the endocochlear potential, together with an endolymphatic potassium concentration of about 150 mmol/l, both values abnormal for an extracellular fluid and crucial for providing the electrochemical driving force that pushes the transduction current through the apical mechanoelectrical channels of the hair cells^[28,29,30]. Since the intracellular potential of the hair cells is negative (about -40 to -70 mV), the total electrical gradient across the apical membrane may exceed 100 to 150 mV, an extraordinarily high value by the standards of sensory synapses, which decisively enhances the sensitivity of auditory transduction even without active amplification^[28,5]. The appearance of the endocochlear potential during postnatal development follows progressive maturation of the stria vascularis, functionally correlated with opening of the ear canal and the appearance of definitive acoustic sensitivity^[34].

3.5 Strial dysfunction: ageing, ototoxicity and clinical implications

The stria vascularis is one of the cochlear structures most vulnerable to physiological ageing and to ototoxic insults. In strial (or metabolic) presbycusis, atrophy of the capillaries and of the marginal and intermediate cells reduces the endocochlear potential, with a consequent loss of gain in the electromotile amplification of the outer hair cells and a hearing loss spread mainly across a wide frequency range, even without direct loss of sensory cells^[1,31]. Studies in the Mongolian gerbil, the reference model for cochlear ageing without noise, have documented a close correlation between the degree of strial atrophy and the extent of the fall in endocochlear potential, although more recent work has shown that inner hair cell synaptopathy is an early degenerative event, partly independent of strial dysfunction^[31,1]. Ototoxicity from platinum-based chemotherapy also primarily involves the stria vascularis, with microvascular damage and degeneration of the marginal cells associated with hair cell death and degeneration of the spiral ganglion, in a multifactorial pathogenetic picture mediated by oxidative stress^[2,1]. Detailed understanding of the cellular and molecular architecture of the stria vascularis therefore remains an essential prerequisite for the pathophysiological interpretation of many forms of acquired and genetic hearing loss, and for the development of therapeutic strategies aimed at restoring cochlear ionic homeostasis^[29,33,32].

4The tectorial membrane

The tectorial membrane (membrana tectoria) is an acellular, gelatinous and highly organised structure overlying the sensory epithelium of the organ of Corti, in anatomical and functional continuity with the hair bundles of the outer hair cells and, more loosely, with those of the inner hair cells. Although it has no cells of its own, its supramolecular architecture is the product of a complex epithelial secretion that takes place in the early stages of cochlear development and that irreversibly sets, in the adult, the viscoelastic properties needed for mechanical coupling with the underlying epithelium ^[3,5]. Understanding its molecular composition and its geometry of attachment is essential for correctly interpreting the biomechanical mechanisms underlying cochlear amplification and frequency tuning, topics that have been the subject of an extensive experimental literature on hair bundle mechanics and outer hair cell electromotility ^[25,26,27].

4.1 Architecture and molecular composition

Ultrastructurally, the tectorial membrane is organised as a fibrillar matrix embedded in a ground substance rich in glycosaminoglycans, within which run radially and longitudinally oriented collagen bundles anchored to a network of non-collagenous glycoproteins. Fibrillar collagens of types II, IX and XI form the load-bearing scaffold of the matrix: type II collagen provides the main fibrillar component, while types IX and XI, arranged heterotypically with it, regulate fibril diameter and three-dimensional organisation, as described for other specialised extracellular matrices of the body. Associated with this collagenous component are glycoproteins specific to the inner ear, α- and β-tectorin and otogelin, which crosslink the matrix and give it the anisotropic viscoelastic properties indispensable for the selective transmission of mechanical forces to the hair bundles. Although this composite architecture has no transducing activity itself, it directly conditions the biomechanics of the stereociliary complex described in the classical studies of the organisation of the organ of Corti, in which the gradation of stereocilia height along the cochlear axis is identified as the main anatomical correlate of tonotopic tuning ^[3]. The radial stiffness of the membrane, greater than its longitudinal stiffness, allows localised propagation of deflection waves, a mechanical requirement consistent with models of cochlear amplification based on local interactions between the supracellular structure and active force generators ^[23,24,27].

4.2 Zone of attachment and relations with the limbus spiralis

Medially the tectorial membrane arises from the limbus spiralis, anchored to the surface of the limbal lip by a thickened region known as the primary attachment zone, rich in collagen fibrils inserting deeply into the interdental matrix. This region gives mechanical stability to the medial pole of the structure, acting as a fulcrum about which the membrane makes its radial movements during deflection of the cochlear partition. Laterally the thinned tectorial membrane extends to contact, or nearly contact depending on the tonotopic region and the species, the surface of the Hensen cells, establishing a second mechanical constraint less rigid than the limbal one. Between these two ends, the undersurface of the membrane bears a reticular imprint corresponding to the layout of the reticular lamina and to the insertion points of the tallest stereocilia of the outer hair cells, a configuration reflecting the close morphofunctional interdependence between the sensory epithelium, the supporting pillar and Deiters cells, and the overlying structure ^[9,10]. It has been shown that the integrity of the cytoskeleton of the supporting cells, in particular the organisation of microtubule bundles mediated by regulatory proteins such as GAS2, is decisive for maintaining the correct geometry of the cochlear partition and, indirectly, for efficient transmission of sound energy to the tectorial-stereociliary complex ^[10,11]. Studies of selective ablation of pillar and Deiters cells have also confirmed that loss of this supporting scaffold severely impairs postnatal development of the cochlea and hearing, underlining that the function of the tectorial membrane cannot be separated from the structural context in which it operates ^[9].

4.3 Role in stimulating the outer hair cells

The primary functional role of the tectorial membrane is direct mechanical coupling with the tallest stereocilia of the outer hair cells, which contain the mechanoelectrical transduction channels activated by tension in the apical links (tip links) ^[26]. As the travelling wave moves along the basilar membrane, radial shearing between the reticular lamina and the undersurface of the tectorial membrane deflects the hair bundles, opening the transduction channels and causing receptor depolarisation ^[25,26]. This bidirectional mechanical coupling is central to models of the so-called cochlear amplifier, according to which the force generated by the somatic electromotility of the outer hair cells, mediated by the motor protein prestin, is fed back into the system through the tectorial membrane itself, creating a positive feedback loop able to amplify vibrations selectively at the characteristic region for each frequency ^{[18,19,22,24]}. In vivo vibrometric measurements have shown that differential motion between the apical and basal poles of the outer hair cells, consistent with functionally effective electromotility even at the highest frequencies, results in cycle-by-cycle amplification of the motion of the cochlear partition, in which the tectorial membrane acts as a transmission element and as a reactive mechanical load ^[17,21]. Alternative global hydromechanical models also attribute to the stiffness and mass of the tectorial membrane an active role in generating a longitudinal pressure wave that contributes, together with the local component, to the sensitivity and frequency selectivity of the whole system ^[23]. It should also be noted that contractile activity of the non-sensory supporting cells, such as that mediated by TRPA1 channels in the Hensen cells after acoustic trauma, can dynamically modulate the geometry of the subtectorial space and hence the efficiency of mechanical coupling between the tectorial membrane and the hair bundles, a homeostatic mechanism regulating cochlear sensitivity ^[13]. Finally, cholinergic efferent control of the outer hair cells, mediated by the α9α10 nicotinic receptor, acts indirectly on the dynamics of the whole tectorial membrane and hair bundle complex by modulating somatic stiffness and hence the transmission of locally generated forces, integrating the physiology of the tectorial membrane into the wider neural feedback circuit of the cochlea ^[4,36].

5The spiral ganglion

5.1 Rosenthal’s canal and topographic organisation

The spiral ganglion of Corti occupies Rosenthal’s canal, a spiral bony cavity hollowed out in the modiolus that follows the two and a half turns of the human cochlea along their whole extent, lying immediately medial to the osseous spiral lamina and in close contact with the cochlear bundle of the acoustic nerve^[41,5]. Within this canal lie the cell bodies of the primary afferent neurons, whose peripheral processes pass through the canaliculi of the osseous spiral lamina (the so-called canales nervi cochleae) to reach the sensory epithelium of the organ of Corti, while their central processes join the trunk of the cochlear nerve running to the cochlear nuclei of the brainstem^[41,6]. The topographic distribution of the cell bodies in the canal maintains a precise tonotopic organisation mirroring that of the basilar membrane: neurons destined for the basal regions, serving high frequencies, occupy the basal parts of the canal, while those destined for the apex, corresponding to low frequencies, lie in its most apical parts, preserving the frequency map along the whole course of the acoustic nerve^[5,41].

5.2 Type I afferent neurons

The neuronal population of the spiral ganglion is functionally and morphologically heterogeneous. Type I neurons make up about 90 to 95% of the total and are bipolar, with a large cell body surrounded by a continuous myelin sheath extending along both the peripheral and central processes^[41,40]. Each type I neuron makes a single, uniform synaptic contact with one inner hair cell (IHC), an anatomical and functional relationship that reflects the role of the inner hair cell as the true primary mechanoelectrical transducer of the acoustic signal^[5,3]. The afferent synapse between the IHC and the dendrite of the type I neuron is glutamatergic, organised around presynaptic ribbon synapses that allow rapid, synchronous and sustained vesicle release, indispensable for fine temporal coding of sound^[6,8]. The role of this synaptic architecture is clear in mouse models lacking otoferlin, the vesicle fusion protein essential for glutamate release at the IHC synapse: its absence abolishes synaptic transmission even though the hair cells are structurally intact, and is associated with distress and selective apoptosis of the type I neurons of the spiral ganglion, confirming the close trophic and functional dependence of these neurons on peripheral synaptic activity^[37].

5.3 Type II afferent neurons

Type II neurons, which make up the remaining 5 to 10% of the ganglion population, have clearly distinct morphological and molecular features. They are pseudounipolar neurons with smaller cell bodies and unmyelinated peripheral processes that cross the tunnel of Corti transversely to reach the row of outer hair cells (OHCs), where each type II neuron, running spirally towards the base, innervates a large number of cells belonging to several rows^[41,42]. This population has historically been identified by immunohistochemistry for peripherin, a type III intermediate filament expressed selectively by type II neurons from the prenatal stages of cochlear development, with an expression pattern that strengthens progressively in the basal turns during ontogeny^[43,40]. The synapses between OHCs and type II neurons differ qualitatively from type I synapses: they are weaker glutamatergic synapses, with a lower density of active sites and less effective transmission, consistent with the hypothesis that type II neurons respond preferentially to intense, potentially damaging sound rather than taking part in the ordinary coding of auditory information^[36,38]. Also relevant is the discovery of a direct GABAergic efferent innervation of type II afferent neurons from medial olivocochlear neurons, which forms a complex local circuit able to modulate the activation threshold of this nociceptive-like pathway before the signal even reaches the brainstem^[36,38]. A further molecular characteristic is the persistent expression of the purinergic P2X1 receptor in the type II neurons of the adult rat, a finding that suggests a prolonged role for ATP-mediated signalling in this subpopulation well beyond the stages of postnatal development^[39].

5.4 Convergence and divergence

The numerical relation between hair cells and primary afferent neurons directly reflects the functional asymmetry between the inner and outer pathways. Each IHC normally receives 10 to 20 independent dendritic terminals from as many type I neurons, each of which projects exclusively to that single cell, creating a pattern of divergence from the sensory cell to the ganglion that multiplies the parallel channels of acoustic information and allows their diversification in threshold and spontaneous firing^[5,3]. The OHC pathway, by contrast, is dominated by convergence: a single type II neuron gathers input from a large number of outer hair cells, often belonging to different rows and different segments of the cochlear turn, creating a sparse and integrative rather than point-to-point synaptic architecture^[41,42]. This quantitative and organisational disparity, many neurons for each IHC against few neurons summing input from many OHCs, mirrors the numerical relation between the two hair cell populations themselves and is already established in early postnatal development, when a selective loss of about a quarter of the total population, mainly affecting type II neurons, occurs during the first postnatal week in the mouse, a refinement process that definitively sculpts the convergence relations seen in the adult^[40].

5.5 Periphery, centralisation and functional integration

The peripheral process of the type I neurons crosses the osseous spiral lamina radially, maintaining a spatial segregation that is preserved in their central projection to the cochlear nucleus, where peripheral tonotopy is faithfully reconstituted along a dorsoventral axis, the basis of the so-called tonotopic centralisation of auditory information^[5,41]. Type II neurons, by contrast, run a spiral basal course before reaching the bony lamina, an anatomical detail that complicates their electrophysiological study in vivo and underlies the relative scarcity of functional data compared with their type I counterparts, although selective labelling with peripherin and optogenetic approaches have recently allowed their synaptic transmission properties and associated efferent circuits to be defined more precisely^[42,38]. Overall, the organisation of the spiral ganglion, with its dedicated bony canal, its type I and type II neuronal duality, its asymmetric convergence and divergence and its glutamatergic ribbon synapses of different strength, forms the anatomical infrastructure that translates the very rich information generated by the organ of Corti into an ordered, hierarchical and tonotopically coherent neural code, preparing the central processing of the auditory signal^[6,8,41].

6The vestibular epithelia

6.1 General architecture of the maculae and cristae

The peripheral vestibular apparatus shares with the cochlea the fundamental cellular principle of the mechanosensory receptor, the hair cell, but organises it into five distinct sensory epithelia: the two maculae of the otolith organs, the macula utriculi and macula sacculi, and the three cristae ampullares of the semicircular canals. All these epithelia, although specialised to detect linear accelerations (including gravitational and inertial acceleration) and angular accelerations respectively, reproduce a common histological pattern consisting of a mosaic of hair cells interspersed with supporting cells, topped by an acellular accessory structure that couples the stimulus mechanically to the hair bundle^[44,50]. As in the organ of Corti, vestibular mechanoelectrical transduction depends on a bundle of stereocilia arranged in order of increasing height, anchored by apical links (tip links) that open mechanotransduction channels in response to submicrometre deflections^[3,26,25]. Unlike the cochlea, however, the vestibular epithelia contain two morphologically and functionally distinct populations of hair cells, called type I and type II, whose coexistence is a feature exclusive to amniotes^[44,47].

6.2 Type I and type II hair cells

The type I hair cell is flask-shaped, with a narrow neck and a basolateral body almost entirely enveloped by a single cup-shaped afferent ending, the calyx, which surrounds the cell up to its apical pole, leaving only the hair region exposed^[44,47,49]. This unique synaptic configuration gives the type I cell peculiar electrophysiological properties, including a low-threshold potassium current, I_K,L, responsible for a low resting membrane resistance and a non-linear, high-gain response dynamic^[45,49]. In the narrow synaptic space of the calyx, accumulation of potassium ions released by the receptor cell during transduction contributes a non-quantal component of transmission, distinct from classical vesicular glutamate release, able to depolarise the calyceal membrane directly through changes in extracellular ion concentration in the synaptic cleft^[45,46]. This hybrid transmission, both quantal and non-quantal, allows a speed of signal transfer unmatched in the entire sensory system, consistent with the minimal latency required by the vestibulo-ocular reflexes^[46,50].

The type II hair cell, by contrast, keeps a more traditional cylindrical shape, similar to that of the cochlear hair cells, and makes multiple synaptic contacts with several afferent bouton endings, each of which may receive input from several cells^[47,48]. Synaptic transmission at the type II cell is entirely quantal and glutamatergic, mediated by postsynaptic AMPA receptors, with slower kinetics than the non-quantal component of the calyx^[48,6]. Both cell populations receive efferent innervation from the brainstem vestibular efferent system: in type II cells cholinergic efferent terminals act directly on the receptor cell through α9-containing nicotinic receptors coupled to small-conductance calcium-activated potassium (SK) channels, causing hyperpolarisation and reduced afferent sensitivity^[51,48]; in type I cells the efferent action is instead mainly indirect, mediated by synapses on the calyceal afferent fibres themselves, including GABAergic components also described in the cochlea for the type II afferents of the outer hair cells, evidence of a logic of efferent modulation shared by the two labyrinths^[36,38].

6.3 Supporting cells and epithelial homeostasis

The vestibular supporting cells, arranged in columns between the hair cell bodies and anchored both to the basal lamina and to the apical reticular surface, perform functions comparable to those of the cochlear supporting cells: maintaining the structural integrity of the epithelium, regulating the ionic microenvironment around the hairs and possibly sequestering the potassium released by the sensory cells during transduction, in an ion recycling circuit similar to that well described for the organ of Corti^[32,33]. The cytoskeleton of these cells, rich in microtubules and actin filaments organised into rigid supporting structures, ensures the mechanical cohesion of the sensory epithelium during stimulus-induced deformation, a principle of cytoskeletal specialisation of supporting cells that has a direct parallel in the function of the pillar and Deiters cells of the cochlea^[10,9].

6.4 The otolithic membrane, otoconia and cupula

In the otolith maculae, the stereociliary bundles and the kinocilium project into an acellular gelatinous matrix, the otolithic membrane, on whose surface lies a crystalline layer of otoconia, calcium carbonate concretions in an organic matrix whose density, greater than that of the endolymph, gives the whole complex enough inertia to act as a seismic mass for detecting linear acceleration and gravity^[44,50]. In the ampullary cristae the homologous accessory structure is the cupula, a gelatinous membrane without otoconia that extends from the crista to the roof of the ampulla, occluding it completely: endolymph displacement caused by angular acceleration deflects the cupula and with it the hair bundles embedded in it, through a purely hydrodynamic coupling mechanism^[50].

6.5 The striola and morphological polarisation

A distinctive feature of the otolith maculae is the striola, a curved band that crosses the macula and divides it into two regions with opposite hair bundle orientation. The morphological polarisation of each cell is defined by the position of the kinocilium relative to the graded row of stereocilia: in the utricular macula the cells are oriented with the kinocilium facing the striola, while in the saccular macula the polarity is reversed, with the kinocilium lying on the far side of the striola^[3,44]. This systematic reversal of orientation, known as the line of polarity reversal (LPR), creates within each organ a population of receptors with mutually opposite vectors of maximum sensitivity, a condition that allows bidirectional coding of accelerations along the same functional axis and gives the striolar regions distinct dynamic properties, with type I cells predominating and faster response kinetics than the extrastriolar periphery, dominated by type II cells with more tonic responses^[49,47]. The regional distribution of the two cell types and the corresponding arrangement of calyces and synaptic boutons therefore form a system of parallel and complementary coding, in which the cellular topography itself is the anatomical substrate of directional discrimination of the vestibular stimulus^[50,48].

7Clinical and experimental correlates

The microanatomical description of the cochlea takes on its full clinical meaning only when set in constant dialogue with human histopathology, with experimental animal models and with the imaging techniques that have made its characterisation possible. Presbycusis, drug ototoxicity, cochlear synaptopathy and genetic deafness share a common cellular substrate, namely hair cells, supporting cells, stria vascularis and spiral ganglion neurons, whose selective or combined impairment explains the extraordinary variety of audiological phenotypes seen in clinical practice^[1,5,31].

7.1 Histopathology of the human cochlea in ageing and ototoxicity

Human temporal bone studies and mouse models agree in identifying the stria vascularis and its microvascular bed as one of the earliest and most vulnerable targets of cochlear ageing, with a consequent fall in endocochlear potential and collapse of the electromotile amplification of the outer hair cells, the mechanism underlying the high-frequency hearing loss typical of presbycusis^[1,28,29]. In parallel, work on the Mongolian gerbil under conditions of quiet ageing has made it possible to separate the contribution of strial dysfunction from the earlier contribution of cochlear synaptopathy, the selective loss of synapses between inner hair cells and type I afferent fibres, a degenerative event that precedes the death of the hair cells themselves and escapes conventional audiometric methods^[31,6]. On the iatrogenic side, cisplatin ototoxicity is the most studied experimental paradigm: the drug enters the cells of the inner ear through specific transport channels, induces mitochondrial oxidative stress and leads to hair cell death, degeneration of the spiral ganglion and structural damage to the stria vascularis, in a pathogenetic sequence that recapitulates in accelerated form many of the events of physiological ageing^[2,1]. The phenotypic convergence between ototoxicity and presbycusis has important diagnostic implications, since both present clinically with a threshold rise predominantly at high frequencies, reflecting the topographic vulnerability of the cochlear base, where the hair cells that transduce high tones lie^[3,2].

7.2 Animal models and the transferability of experimental data to humans

Almost all our knowledge of cochlear cell biology derives from genetically modified mouse models, in which conditional ablation or ectopic expression of specific genes has allowed precise functions to be attributed to individual cell populations. Selective ablation of pillar and Deiters cells with the Cre/loxP system showed that primary loss of supporting cells severely impairs postnatal development of the organ of Corti and hearing, even without direct damage to the hair cells, confirming the irreplaceable structural role of these cells^[9,7]. Similarly, mice mutant for the GAS2 gene show disorganised microtubule bundles in the pillar and Deiters cells with consequent hearing loss, a phenotype faithfully reproduced in humans carrying orthologous mutations, evidence of the high phylogenetic conservation of cochlear cytoskeletal architecture between rodents and humans^[10]. Prestin knockout mice, as well as mice expressing non-motile prestin, have allowed the role of somatic electromotility to be separated experimentally from the purely structural role of the protein, clarifying the relative contribution of each mechanism to cochlear amplification in vivo^[18,22,16]. The Otof knockout model, which reproduces human auditory neuropathy due to otoferlin mutation, showed selective apoptosis of type I ganglion neurons in the presence of preserved peripheral cochlear responses, a histopathological correlate that explains the clinical dissociation between normal otoacoustic emissions and absent brainstem auditory responses typical of this condition^[37,41]. Optogenetics applied to supporting cells is a more recent methodological development, which has shown direct control by the Deiters and outer pillar cells over the cochlear electromechanical response, opening prospects for selective non-pharmacological manipulation of peripheral auditory function^[12,13].

7.3 Investigative techniques: immunohistochemistry, confocal and electron microscopy

The development of immunohistochemistry applied to the cochlea, beginning with the pioneering work of the 1980s, has allowed precise localisation of neurotransmitters, receptors and structural proteins within the organ of Corti, overcoming the limits of conventional light microscopy alone^[8,4]. Transmission electron microscopy remains irreplaceable for the ultrastructural characterisation of the afferent synapses between inner hair cells and ganglion dendrites, whose distinctive morphology, with presynaptic ribbons, postsynaptic densities and organised synaptic vesicles, allows a correlated functional and structural analysis that no other method can reproduce^[6]. Cell-specific transcriptomic analysis applied separately to adult pillar and Deiters cells has revealed expression of genes encoding molecular machinery characteristic of hair cell specialisations, redefining the functional boundary between sensory and supporting cells and suggesting a latent regenerative potential relevant to future therapeutic strategies^[15,14,11]. Finally, single-particle cryo-electron microscopy of dolphin prestin has clarified at atomic resolution the conformational cycle underlying electromotility, linking cell physiology with molecular structural biology in an interpretive continuum from molecule to organ^[19,17].

7.4 Clinical relevance for the diagnosis and treatment of hearing loss

The integration of histopathological, molecular and ultrastructural data now provides a conceptual framework that guides clinical practice: the distinction between deafness of strial, synaptic, neuronal or supporting cell cytoskeletal origin guides the choice of targeted diagnostic strategies (otoacoustic emissions, auditory evoked potentials, genetic testing) and treatments, including the prospect of gene therapy for monogenic forms such as those due to mutations of GAS2 or otoferlin^[10,37,33]. The emerging role of the non-sensory supporting cells, including the Hensen cells, which carry TRPA1 channels activatable by products of oxidative stress after acoustic trauma, also suggests new pharmacological targets for cochlear protection after noise exposure, widening the therapeutic horizon beyond the hair cell alone^[13,9].

1General framework: the multi-tissue origin of the ear

Embryologically, the auditory and vestibular apparatus is one of the anatomical systems with the most complex tissue origin in the body. None of the three classically distinguished portions, external, middle and inner ear, derives from a single germ layer or a single progenitor cell population: their morphogenesis results instead from the spatiotemporal interweaving of at least four distinct cell contingents, namely the surface ectoderm specified to form the otic placode, the cells of the cranial neural crest, the endoderm of the pharyngeal pouches and the paraxial and neural crest mesenchyme that populates the pharyngeal arches^[6,15,31]. This composite origin explains why ear malformations in clinical practice almost never respect traditional anatomical boundaries but present as syndromes simultaneously involving apparently distant structures such as the kidney, the mandible or the heart^[5,35].

1.1 The otic placode as the autonomous matrix of the inner ear

The inner ear arises from a focal thickening of the cranial ectoderm, the otic placode, which appears bilaterally next to rhombomeres 5 and 6 of the developing hindbrain^[29,32]. Unlike the middle and external ear, the otic vesicle derived from it has considerable developmental autonomy: heterotopic transplants show that, once induced, it can carry out much of its morphogenetic programme even without the surrounding tissues that determined its initial induction^[6,7]. Induction of the placode depends on diffusible signals from the paraxial mesoderm, the notochord and the hindbrain itself, among which members of the fibroblast growth factor family (Fgf3, Fgf8, Fgf10) play a decisive and partly redundant role, as shown by combined knockdown studies in zebrafish and mouse^[9,12,13]. These are joined by Wnt signals and by regulation mediated by cell adhesion molecules (N-CAM, L-CAM, Ng-CAM) that define early the boundaries between neural and non-neural placodal territory^[10,11]. It is worth stressing that, although the otic vesicle has historically been described as a purely ectodermal derivative, recent evidence in the mouse shows a direct, albeit minor, contribution of neuroepithelial cells from the neural tube, including neural crest cells, to the vesicle itself: a finding that redefines the notion of an exclusively placodal origin of the inner ear and underlines its dual nature^[8].

1.2 The cranial neural crest and the pharyngeal arches: the matrix of the middle and external ear

While the inner ear has its roots in placodal ectoderm, the ossicular chain, most of the tympanic cavity and the cartilaginous structures of the auricle and external auditory canal derive from cranial neural crest cells that have migrated into the first and second pharyngeal arches^[3,19]. The first (mandibular) arch classically contributes the malleus and incus, while the second (hyoid) arch is held responsible for forming the stapes, although whether the manubrium of the malleus and the long process of the incus have a single- or double-arch origin remains debated, the classical otolaryngological literature proposing a double-arch interpretation not fully confirmed by more recent embryological studies in human and mouse models^[3]. These ectomesenchymal cells undergo chondrogenesis and subsequent endochondral ossification under the coordinated control of secreted signals, Sonic hedgehog, bone morphogenetic proteins (BMPs) and arch-specific homeotic transcription factors, whose disruption produces the spectrum of isolated or syndromic ossicular malformations^[19,20]. The external auditory canal in turn develops from the first pharyngeal (branchial) groove, whose incomplete canalisation or developmental arrest produces the spectrum of congenital atresia and stenosis, almost always associated with anomalies of the tympanic cavity and ossicular chain, since both share the same arch origin^[4,1]. The most severe syndromic forms of first and second arch dysgenesis, as in otocephaly, illustrate dramatically the close interdependence of auricular, mandibular and lingual structures sharing an origin from the cranial neural crest^[2].

1.3 The endodermal contribution: the pharyngeal pouches

The tympanic cavity and the Eustachian tube derive from expansion of the first endodermal pharyngeal pouch, whose epithelium lines the whole tympanic cavity and secondarily surrounds the ossicles of ectomesenchymal origin^[1,19]. The interaction between the endodermal epithelium of the pouch and the surrounding neural crest ectomesenchyme is bidirectional and necessary for correct pneumatisation of the tympanic cavity and for separation of the ossicles from the undifferentiated mesenchymal blastema; a breakdown in this epithelial-mesenchymal dialogue is implicated in the genesis of isolated ossicular malformations, a condition that is in any case rare compared with combined syndromic forms^[1]. The EYA1 gene, mutated in branchio-oto-renal syndrome, is also expressed in the otic epithelium and in the branchial and renal structures, evidence of a genetic network shared by apparently heterogeneous organs united by a common developmental programme mediated by pleiotropic genes^[5].

1.4 Mesenchyme, integrated chronology and clinical significance of the composite origin

The periotic mesenchyme, of mixed mesodermal and neural crest derivation, plays a permissive and inductive role towards both the otic vesicle and the cartilaginous otic capsule, providing a microenvironment of retinoic acid, Shh and BMP signals essential to the three-dimensional morphogenesis of the labyrinth^[16,18,20]. These four contingents, placode, neural crest, endoderm and mesenchyme, do not develop in a rigidly compartmentalised sequence but within a broadly overlapping time window, from the fourth to the tenth week of gestation, with intense molecular crosstalk that makes the whole system extremely vulnerable to pleiotropic genetic or teratogenic insults, as shown by CHD7 mutations in CHARGE syndrome, which simultaneously disrupt development of the otic vesicle, the cochlea and non-auditory craniofacial structures^[35]. A fundamental clinical corollary follows, stressed by the paediatric neuroradiological literature: almost all malformations of the external auditory canal are accompanied by anomalies of the tympanic cavity and ossicles, while isolated ossicular malformations not associated with first or second arch anomalies are rare^[1,4]. Understanding the multi-tissue origin of the ear is therefore not merely an exercise in descriptive embryology but the indispensable conceptual foundation for correctly interpreting the radiological semiology of congenital malformations, anticipating the syndromic associations expected in a given patient and rationally planning the reconstructive surgical approach, which must take into account the distinct embryological origin, and hence the distinct genetic vulnerability, of each component of the auditory and vestibular system^[1,3,19].

2Development of the external and middle ear

The external and middle ear derive from a complex embryological mosaic in which the first branchial arch (mandibular) and the second branchial arch (hyoid), together with the first pharyngeal pouch and the first ectodermal groove, combine to form the auricle, external auditory canal, ossicular chain and tympanic cavity ^[4][1]. Understanding this morphogenetic sequence is essential for correctly interpreting the spectrum of congenital malformations of the external and middle ear, whose severity is directly proportional to the timing and extent of the developmental arrest ^[4][1].

2.1 The hillocks of His and the morphogenesis of the auricle

The auricle arises from the fifth to sixth week of gestation from the proliferation of six mesenchymal swellings arranged around the first branchial groove, the colliculi auriculares described by His: three of mandibular (first arch) and three of hyoid (second arch) derivation ^[4]. Asynchronous fusion and differential growth of these primordia determine the final topography of the tragus, helix, antihelix and lobule; asymmetries or arrests at this stage produce the spectrum of microtia and atresia, often associated with syndromic involvement of the second arch ^[1][4]. It is significant that developmental disorders of the first two branchial arches may be associated with renal and cervical malformations in branchio-oto-renal syndrome, in which the EYA1 gene pleiotropically orchestrates the morphogenesis of structures embryologically distant yet sharing an origin from cranial placodes and mesenchymal derivatives ^[5].

2.2 The external auditory canal and the first branchial groove

The external auditory canal arises from the ectodermal invagination of the first branchial groove, which advances medially until it meets the endodermal outpouching of the first pharyngeal pouch, from which it remains separated by a thin three-layered meatotympanic membrane ^[4][1]. The most medial part of the groove gives rise, through secondary canalisation of a solid epithelial plug (the meatal plug), to the definitive lumen of the canal, an event completed only around the twenty-first week of intrauterine life; an arrest at this stage of recanalisation underlies congenital atresia and stenosis of the external auditory canal, frequently accompanied by anomalies of the ossicular chain because of their common origin from the first and second arches ^[4][1]. The clinical-radiological correlation between the degree of bony atresia, the state of the ossicular chain and mastoid pneumatisation remains the basis of preoperative prognostic classification in paediatric conductive hearing loss ^[1].

2.3 Meckel’s and Reichert’s cartilages and the origin of the ossicular chain

The mesenchyme of the first arch condenses into Meckel’s cartilage, whose dorsal pole gives rise to the malleus and incus, while the mesenchyme of the second arch forms Reichert’s cartilage, from which the stapes superstructure derives ^[3][19]. Recent three-dimensional reconstruction studies of human embryos have, however, challenged the dogma of an ossicular origin entirely from the first arch, showing that the malleus and incus form from distinct chondrification centres that later fuse, with mixed mesenchymal contributions whose exact arch origin remains debated ^[3]. The stapes footplate, by contrast, retains a composite origin, deriving partly from the otic capsule itself rather than exclusively from the second arch, which explains the relative sparing of the oval window in some forms of isolated ossicular aplasia ^[3][19]. Ossicular morphogenesis is governed by a network of secreted signals and transcription factors, including members of the BMP family, Wnt pathways and pharyngeal arch-specific homeotic regulators, which coordinate chondrification, the formation of the incudomalleolar and incudostapedial joints and subsequent endochondral ossification ^[19][20].

2.4 The first pharyngeal pouch, the Eustachian tube and the tympanic cavity

The first pharyngeal pouch evaginates laterally to form the recessus tubotympanicus, whose narrow proximal part becomes the Eustachian tube and whose distal part, as it expands, gives rise to the primitive tympanic cavity ^[4][19]. It is the progressive expansion of this endodermal cavity that pneumatises the ossicular chain, previously free in the connective mesenchyme, and covers it with mucosa, a process that proceeds from lateral to medial and that, if incomplete, leaves the ossicles embedded in persistent connective tissue responsible for congenital ankylosis independent of canal malformations ^[4][1]. Dysgenesis of the pharyngeal pouch is also part of a wider spectrum of anomalies of the branchial apparatus, as illustrated by experimental models of first and second arch dysplasia in which ventromedial displacement of the auricular structures accompanies major craniofacial defects ^[2].

2.5 The three-layered tympanic membrane

The tympanic membrane forms at the interface between the three germ components involved in middle ear development: an outer layer of ectodermal derivation, continuous with the epithelium of the external auditory canal, an intermediate fibrous layer of mesenchymal origin (from the mesoderm of the first and second arches), and an inner mucosal layer of endodermal derivation, continuous with the mucosa of the tympanic cavity derived from the first pharyngeal pouch ^[4][1]. This three-layered architecture, which faithfully recapitulates the triple germ origin of the whole tympano-ossicular complex, explains the selective vulnerability of the intermediate fibrous layer in traumatic perforations and the relative preservation of the epithelial lining layers, of crucial importance for reconstructive myringoplasty ^[4]. The integration of these events, the hillocks of His, canalisation of the canal, chondrification of Meckel’s and Reichert’s cartilages, expansion of the tubotympanic recess and layering of the tympanic membrane, thus forms a single morphogenetic continuum whose interruption at any level produces the wide clinical spectrum of congenital conductive hearing loss in children, a constant subject of diagnostic and radiological assessment and personalised surgical planning ^[1][5][19].

3The otic placode and the otocyst

The inner ear arises from a transient embryonic structure, the otic placode, a thickening of the cephalic ectoderm that is the common progenitor of all the sensory and non-sensory components of the membranous labyrinth^[6,15,31]. Studies in avian, mouse and zebrafish models have made it possible to reconstruct in considerable detail the molecular and morphogenetic steps leading from specification of the preplacode to formation of the otocyst, revealing at the same time a striking phylogenetic conservation of the regulatory networks involved^[7,29]. A dual contribution to the origin of the otic vesicle has also emerged, with cell populations coming both from the otic placode proper and from adjacent ectodermal territories, suggesting a greater histogenetic complexity than the classical single-phase model^[8].

3.1 The preplacodal region and inductive signals

Formation of the otic placode is preceded by the establishment of a band of competent cephalic ectoderm, the pre-placodal region, which surrounds the neural plate and gives rise to all the sensory cranial placodes^[9,15]. Within this territory, otic induction proper requires the coordinated action of fibroblast growth factors secreted by the paraxial mesoderm and the developing hindbrain, in particular Fgf3 and Fgf10, whose combined expression is necessary and partly redundant for initiating the placode in mammals^[13]. Gene deletion studies in the mouse have shown that simultaneous inactivation of Fgf3 and Fgf8 drastically impairs both induction and maintenance of the placode and the subsequent vesicle, indicating a joint requirement for these two isoforms^[12]. In zebrafish, similarly, the combined activity of Fgf8 and Fgf3 governs not only initial induction but also the subsequent regional patterning of the inner ear, confirming the universality of the FGF-dependent circuit across the vertebrate phylogenetic scale^[9]. The general importance of the FGF signalling pathway in otic development, including its later branches in hair cell morphogenesis, has been the subject of systematic reviews stressing its centrality throughout the ontogeny of the organ^[11,33].

3.2 Adhesion molecules and the delimitation of placodal boundaries

Alongside diffusible inductive signals, the establishment of sharp boundaries between the neural ectoderm, the otic placode and the surrounding epidermal ectoderm depends on differential expression of cell adhesion molecules, which give the placodal epithelium cohesive properties distinct from those of the adjacent tissues^[10]. This early compartmentalisation is the mechanical prerequisite for the focal epithelial thickening that precedes invagination, and is accompanied by changes in the cortical cytoskeleton and in the apicobasal polarity of the placodal cells, elements indispensable for the subsequent bending of the sheet^[31].

3.3 Invagination and closure of the otocyst

In the human embryo, invagination of the otic placode occurs at about the fourth week of gestation, forming an otic pit that progressively deepens and separates from the surface ectoderm to form a closed vesicle, the otocyst or otic vesicle^[32]. Invagination is not simply passive folding but results from a combination of regional changes in cell shape, convergent extension movements and fine regulation of proliferation, whose disruption in experimental models produces malformed or incomplete otocysts^[31]. Programmed cell death has also been shown to play an active rather than merely accessory role at this stage, helping to remodel the placodal epithelium during closure of the vesicle and in the immediately following phases of morphogenesis^[30]. Comparison between model species, from zebrafish to mammals, has revealed both conserved mechanisms and species-specific variations in the kinetics of invagination, particularly in the relative contribution of luminal cavitation compared with classical epithelial folding^[7].

3.4 Axial regionalisation of the otocyst

Once formed, the otocyst is not a homogeneous structure but rapidly undergoes regionalisation along three main axes, dorsoventral, anteroposterior and mediolateral, which prefigures its future division into a dorsal vestibular compartment and a ventral cochlear compartment^[15,31]. Dorsoventral regionalisation is governed to a significant extent by the Sonic Hedgehog signal, secreted by the notochord and the floor plate of the neural tube, whose ventrodorsal gradient is decisive in specifying the cochlear territory as distinct from the vestibular one, as shown by numerous loss- and gain-of-function studies in the mouse^[18]. Signals of the bone morphogenetic protein family also contribute to dorsoventral regionalisation, acting in cooperation and sometimes in antagonism with Shh in defining the sensory and non-sensory territories of the otocyst^[20]. On the anteroposterior side, retinoic acid signalling plays a fundamental patterning role, since local retinoic acid gradients, generated by synthesising and degrading enzymes expressed in complementary domains, orient the axial polarity of the vesicle and influence the position of the future ganglia and semicircular canals^[16]. Together these morphogenetic signals act in a combinatorial rather than strictly hierarchical way, generating a positional grid that is then translated into expression domains of specific transcription factors, responsible for the subsequent regional differentiation of the sensory and non-sensory structures of the inner ear^[15,29]. This early regionalisation is also the substrate on which the later processes of cochlear elongation and coiling are grafted, whose tonotopic architecture originates precisely in the axial domains established at this otocyst stage^[14].

4The specification gene network

Building the inner ear from a simple ectodermal thickening is not a random event but the result of a hierarchical network of transcription factors and morphogenetic signalling axes that overlap in space and time with great precision^[6][15][31]. The otic placode is specified within a wider preplacodal territory, competent to generate cranial sense organs, through the combinatorial action of genes converging on a small core of regulators: Pax2/Pax8, Eya1, Six1, Sox2, Dlx3/5 and Gata3^[8][10][15].

4.1 The Pax2/Pax8 crossroads in placode specification

The paralogous factors Pax2 and Pax8 are activated early in the preotic ectoderm in response to Fgf3 and Fgf10 signals from the hindbrain and paraxial mesoderm^[9][12][13]. Their expression precisely defines the boundary of the future otic placode and governs its invagination to form the otocyst, the structure from which all the epithelial components of the membranous labyrinth derive^[8][32]. Comparative studies in zebrafish have confirmed that combined loss of Fgf signals upstream of Pax2/Pax8 abolishes placodal induction, showing the obligatory nature of this cascade^[7][9]. FGF signalling also acts later in regulating otocyst proliferation and in the morphogenesis of the cochlear duct, evidence of iterative use of the same ligand at different stages of development^[11][33].

4.2 The Eya1-Six1 module and branchio-oto-renal syndrome

Downstream of and in physical cooperation with Pax2/Pax8 operates the Eya1-Six1 complex, a highly conserved transcriptional module that integrates inductive signals and maintains the proliferative competence of placodal cells^[5][15]. Heterozygous mutations of EYA1 in humans cause branchio-oto-renal syndrome, characterised by branchial fistulae, malformations of the external and middle ear and renal dysplasia, evidence of the pleiotropic role shared by the derivatives of the pharyngeal arches^[5][4]. The human phenotype faithfully recapitulates the knockout mouse models, in which absence of Eya1 arrests otocyst development at the simple vesicle stage, without further differentiation into semicircular canals or cochlear duct^[5][6]. This shows that the Eya1-Six1 network is not merely inductive but strictly necessary for subsequent morphogenetic progression^[31][32].

4.3 Sox2 and the specification of proneural and prosensory domains

The factor Sox2, expressed in a restricted domain of the medial wall of the otocyst, marks the proneural territory from which the neuroblasts of the spiral ganglion will arise and, later, the prosensory domains destined to generate hair cells and supporting cells^[17][37][15]. The dose of Sox2 is critical: reduced levels impair neuroblast delamination and alter the subsequent expression of Atoh1 in the prosensory domain, with direct consequences for cochleogenesis^[25][37]. This places Sox2 at the crossroads between the early specification network and the late differentiation programmes of sensory cells, discussed more fully in the section on the organ of Corti^[17][14].

4.4 Dlx3/5 and Gata3 in dorsoventral regionalisation

The homeobox genes Dlx3 and Dlx5, together with the zinc finger factor Gata3, are responsible for the dorsoventral regionalisation of the otocyst, indispensable for separating the vestibular territory, destined for the semicircular canals, utricle and saccule, from the auditory cochlear territory^[31][32][15]. Loss of Dlx5 function in mouse models produces agenesis of the semicircular canals with relative preservation of the cochlear component, a phenotype mirroring the functional asymmetry of the dorsoventral axis imposed upstream by morphogenetic gradients^[31][29]. Gata3, for its part, acts in synergy with Dlx5 to maintain otic identity throughout development, and its disruption shares overlapping phenotypes with defects of the Pax2/Eya1 network, confirming the functional interconnection of these regulators^[19][20].

4.5 Morphogenetic signalling axes: ventral Shh, dorsal Wnt, retinoic acid

Superimposed on the transcription factor network is a triplet of extracellular signalling axes that orient its expression in space. The Sonic Hedgehog signal, secreted by the notochord and the neural floor plate, defines the ventral compartment of the otocyst, promoting formation of the cochlear duct and repressing dorsal vestibular programmes^[18][14][15]. Conversely, the canonical Wnt pathway, active in the roof of the hindbrain, specifies the dorsal territory destined for the vestibular apparatus, interacting directly with Dlx5 and Gata3 in defining compartmental boundaries^[21][31]. The third axis, mediated by retinoic acid, acts as an anteroposterior modulator and as a permissive factor for correct extension of the cochlear duct, integrating with the Shh and Fgf gradients in a non-linear combinatorial network^[16][14]. The dynamic balance between these three axes, rather than their isolated action, determines the final topology of the labyrinth, which explains how point mutations in single nodes of the network can produce extremely heterogeneous malformation spectra, from simple cochlear hypoplasia to the complete labyrinthine aplasia seen in some complex genetic syndromes^[35][31][32].

In summary, the specification gene network of the inner ear should not be seen as a linear sequence of events but as a redundant, interconnected system in which Pax2/Pax8, Eya1-Six1, Sox2, Dlx3/5 and Gata3 are the stable nodes of a circuit whose spatial geometry is imposed by the Shh-Wnt-retinoic acid axes^[6][15][37]. Understanding this molecular architecture now provides the interpretive framework needed to correlate the malformation phenotypes seen in clinical practice with the underlying genetic defects, a theme taken up again in the discussion of congenital malformations of the labyrinth^[35][1].

5Morphogenesis of the labyrinth

The transformation of the otic vesicle into a three-dimensional labyrinthus membranaceus is one of the most elaborate morphogenetic events in vertebrate organogenesis^[6][31]. From an almost spherical vesicular structure, the otocyst undergoes regional epithelial outpouchings, invaginations and fusions that generate, in a strictly coordinated spatiotemporal sequence, the semicircular canals, the endolymphatic duct and sac complex and the cochlear duct^[32][15]. These processes are not merely mechanical: they are guided by overlapping molecular gradients that assign positional identity along the dorsoventral, anteroposterior and mediolateral axes of the vesicle^[8][16].

5.1 Semicircular canals: the fusion plates

The upper vestibular component of the otocyst expands into two flat outpouchings, the vertical and horizontal primordia, each of which develops in its central part a zone of thinned epithelium called the fusion plate^[31][32]. In this region the epithelial cells progressively lose their apicobasal polarity and undergo a process of resorption involving both extracellular matrix remodelling and programmed cell death, with consequent perforation and subsequent fusion of the two opposing epithelial sheets^[30][31]. The result is the transformation of the flat pouch into a hollow ring-shaped duct, while the unresorbed periphery persists as the semicircular canal proper, continuous with the dilated ampulla at one end^[32][6]. Sonic hedgehog (Shh), produced by the notochord and the floor plate of the neural tube, is required for establishing the dorsoventral polarity that precedes formation of the fusion plates, and its impairment results in selective agenesis of the canal compartment^[18][20]. In parallel, bone morphogenetic proteins (BMPs), in particular Bmp4, delimit the territories of the ampullary cristae and help define the margins of the fusion plate itself^[20][15].

5.2 The endolymphatic duct and sac

The first evolutionary and ontogenetic derivative of the otic vesicle is the dorsomedial outpouching that forms the endolymphatic duct, a structure that appears early and serves as an axial reference for all subsequent labyrinthine morphogenesis^[6][29]. The duct lengthens extracranially, ending in the endolymphatic sac, a thin-walled dilatation with functions of ion reabsorption and regulation of endolymph volume and composition^[32][15]. Mouse and avian models indicate that correct specification of the dorsal region of the otocyst, from which the endolymphatic duct arises, depends on the interaction between Wnt signals from the dorsal hindbrain and locally expressed transcription factors such as Dlx5 and Dlx6, whose loss of function is associated with severe malformations not only of the endolymphatic duct but of the entire vestibular compartment^[8][15][31]. Dlx5/6 expression partly overlaps the territories regulated by Eya1 and by the gene network associated with branchio-oto-renal syndrome, underlining how the same transcriptional module is reused to specify distinct dorsal compartments of the otocyst^[5][8].

5.3 The cochlear duct: elongation and spiral coiling

Ventrally the otocyst gives rise to a tubular outpouching that progressively lengthens and coils on itself to form the cochlear spiral, a process completed in humans by the tenth week of gestation with the acquisition of the characteristic two and three quarter turns of the species^[14][32]. Elongation of the cochlear duct is driven by convergent extension cell intercalation movements within the epithelium, a mechanism that requires correct planar cell polarity and is modulated by Fgf signals from the periotic mesenchyme and the epithelium itself^[14][33][9]. Fgf10, together with Fgf3, acts not only in the early induction of the otic placode but also in later phases of elongation, and is required for the duct to reach its final length^[13][12]. Retinoic acid, synthesised by enzymes expressed in the periotic mesenchyme along a basal-apical gradient, helps orient the direction of coiling and establish the apicobasal polarity of the nascent cochlea^[16][14]. The network of molecular guidance cues, including signals of the netrin and semaphorin families, also orients the growth of the axonal processes of spiral ganglion neurons along the elongating axis of the duct, synchronising innervation with epithelial morphogenesis^[36][37].

5.4 Integration of morphogenetic signals

The spatial coherence of the resulting labyrinth depends on the integration, within a narrow time window, of multiple signalling cascades, Fgf, Bmp, Shh, Wnt and retinoic acid, acting combinatorially rather than in isolated sequence^[15][11][20]. Even modest disturbances of this balance, as shown by studies of CHARGE syndrome models linked to CHD7 mutations, result in a spectrum of cochlear and vestibular malformations ranging from complete aplasia to partial dysplasia, confirming the intrinsic fragility of this multifactorial morphogenetic programme^[35][32].

6Hair cell differentiation

The transition from undifferentiated otic epithelium to a mature sensory mosaic is one of the most refined chapters in the morphogenesis of the inner ear, a process that integrates positional signals, cell cycle exit and hierarchical transcriptional networks in a tightly constrained spatiotemporal sequence^[15][31][37]. Within the developing cochlea, specification of hair cells and adjacent supporting cells originates from a transient epithelial territory called the prosensory domain, identifiable along the cochlear duct as it elongates and coils^[14][32].

6.1 The prosensory domain and cell cycle exit

Formation of the prosensory domain is preceded by a synchronised arrest of proliferation mediated by the cyclin-dependent kinase inhibitor p27Kip1 (CDKN1B), whose expression spreads in a wave from the apical to the basal region of the cochlea, delineating in advance the territory destined to generate the organ of Corti^[14][24]. This postmitotic quiescence is not a passive event but a permissive condition necessary for the progenitors to respond subsequently to differentiation signals without aberrant re-entry into the cycle, a principle confirmed by studies of the programmed cell death that sculpts the boundaries of the membranous labyrinth during histogenesis^[30][31]. Concurrent morphogenetic signals, including the Sonic Hedgehog pathway, help orient the proximodistal polarity and prosensory competence of the cochlear epithelium at early stages, acting in synergy with retinoic acid gradients and with the bone morphogenetic proteins (BMPs) already described as pleiotropic regulators of labyrinthine morphogenesis^[16][18][20].

6.2 Notch signalling and lateral inhibition

The classical mechanism that translates the equivalence of prosensory progenitors into a discrete, regular pattern of hair cells alternating with supporting cells is lateral inhibition mediated by the Notch pathway. The ligands Delta-like 1 (Dll1) and Jagged1 (Jag1), dynamically expressed in nascent progenitors, activate the Notch receptor in neighbouring cells, inducing in them expression of the transcriptional repressors Hes1 and Hes5, which in turn repress the hair cell differentiation programme and direct those cells to a supporting phenotype^[21][23][26]. This circuit, first described in invertebrate models and later confirmed in the mammalian cochlear epithelium, explains why pharmacological or genetic suppression of Notch signalling causes overproduction of hair cells at the expense of supporting cells, a phenomenon observable both during development and, potentially, in regenerative settings^[21][23][22]. It is worth stressing that persistently increased Notch signalling in supporting cells is one of the main obstacles to spontaneous hair cell regeneration in adult mammals, a finding that has guided therapeutic strategies of Notch inhibition in regeneration research^[23][34].

6.3 Atoh1: the master gene of hair cell fate

The basic helix-loop-helix (bHLH) transcription factor Atoh1 (formerly known as Math1) is the indispensable and sufficient molecular determinant for inducing the hair cell phenotype, so much so that it has been called the master gene of cochlear sensory differentiation^[17][25][27]. Analysis of lineage tracing lines (fate mapping) based on the Atoh1 locus has shown that the entire progeny of cells transiently expressing this factor contributes both to the hair cell compartment and, to a variable extent, to the supporting cell compartment, indicating a window of cell fate plasticity wider than initially supposed^[25][27]. Three-dimensional live imaging studies have made it possible to visualise in real time the dynamics of Atoh1 induction in the prosensory domain, revealing a wave of activation that proceeds from the medial-basal to the apical-lateral region, with progressive stabilisation of expression in definitive hair cells and its extinction in the surrounding cells destined for support^[27][14]. Efforts at direct reprogramming using the combination of ATOH1, GFI1 and POU4F3 in mature mammalian cells have, however, revealed epigenetic barriers and limits in intercellular signalling that hinder complete and efficient phenotypic conversion, underlining the complexity of the network beyond the single master factor^[28][17].

6.4 The downstream transcriptional network: Sox2, Gfi1, Pou4f3 and Barhl1

Upstream of and in concert with Atoh1 operates Sox2, a high-mobility group (HMG-box) factor expressed throughout the prosensory domain before its restriction, essential for maintaining progenitor competence and for correct induction of Atoh1 itself, in a relationship of mutual, dose-dependent dependence well characterised in hypomorphic mouse models^[17][37]. Downstream of Atoh1 induction, the zinc finger factor Gfi1 is required for the survival and terminal maturation of newly formed hair cells, while the POU-domain factor Pou4f3 (Brn3c) governs further morphological differentiation, including the organisation of the stereociliary bundle, and its loss leads to early degeneration of the hair cells themselves^[17][53][54]. The homeodomain factor Barhl1, expressed later, contributes to the long-term maintenance and survival of hair cells, particularly in the peripheral vestibular and cochlear regions, thus completing a hierarchical cascade, Sox2 then Atoh1 then Gfi1/Pou4f3 then Barhl1, that marks the successive stages from prosensory progenitor to functionally mature sensory cell^[17][53].

6.5 Genesis of the hair cell and supporting cell mosaic

The integrated outcome of these molecular programmes is the characteristic alternating mosaic of inner hair cells, outer hair cells and several populations of supporting cells (pillar, Deiters and Hensen cells) that forms the functional architecture of the organ of Corti^[24][37][42]. This pattern is not merely histological but underlies the tonotopic organisation of the cochlea and its subsequent orderly innervation by spiral ganglion fibres, whose directed growth depends in turn on the correct spatial placement of the newly differentiated hair cells^[37][47]. Genetic alterations of these pathways, including mutations in genes encoding unconventional myosins and stereociliary bundle proteins, account for many forms of hereditary sensorineural hearing loss, linking developmental biology directly to clinical audiological practice^[49][52][54]. A thorough understanding of this cascade also remains the rational foundation of hair cell regeneration strategies, since controlled recapitulation of the Notch-Atoh1-Sox2 sequence in adult tissue is the primary goal of translational research in regenerative otology^[22][28][34].

7The spiral ganglion: neurogenesis and connections

The spiral ganglion of Corti is the primary neuronal station of the peripheral auditory pathway, and its genesis is one of the earliest determined events in the morphogenesis of the human inner ear ^[8,37]. Its neuronal population arises from a proneural domain of the anteroventral wall of the otic vesicle, distinct from but adjacent to the territory that will give rise to the cochlear sensory epithelium, in keeping with the principle of the dual origin of the membranous labyrinth described in mouse models and later extrapolated to human embryology ^[8,6].

7.1 Neuroblast delamination and early specification

Formation of the cochleovestibular ganglion begins with delamination of neuroblasts from the otic epithelium into the underlying periotic mesenchyme, a phenomenon documentable in the human embryo from the fifth week of development and completed, for the cochlear component, by the eighth week ^[39,32]. This migratory process depends on expression of the basic helix-loop-helix transcription factor Neurogenin1 (Neurog1), whose expression domain in the otic placode and vesicle defines with considerable spatial precision the proneural territory destined to generate the ganglion neuroblasts ^[37,17]. Downstream of Neurog1, activation of the factor NeuroD1 is required for the survival of delaminated neuroblasts and their progressive postmitotic differentiation, in a regulatory circuit that recapitulates, with tissue-specific variations, the proneural neurogenesis programmes described in other regions of the cranial peripheral nervous system ^[37,26]. Notch signalling, through lateral inhibition, helps restrict the pool of proneural progenitors and prevent neuronal overproduction, and its modulation governs the balance between neuronal, sensory cell and supporting cell fates within the adjacent otic domains ^[26,21].

7.2 Ganglion morphogenesis and programmed cell death

Once delaminated, the neuroblasts aggregate near the base of the forming cochlea, progressively organising into the future spiral arrangement that accompanies elongation and coiling of the cochlear duct ^[39,14]. This morphogenetic process is not merely passive but requires precise coordination with the physiological apoptosis of supernumerary neuronal subpopulations, a phenomenon of programmed cell death that sculpts the definitive architecture of the ganglion, much as it regulates the size of the entire membranous labyrinth ^[30,31]. The proportion of neurons that survive depends closely on the local availability of trophic factors produced by the target sensory epithelium, according to a classical principle of neurotrophic competition that takes on particularly strict significance in the inner ear given the narrow time window in which definitive innervation is established ^[38,15].

7.3 Axon guidance and neurotrophic support

The centripetal and centrifugal growth of the processes of ganglion neurons, towards the brainstem and towards the organ of Corti respectively, is governed by an integrated repertoire of axon guidance signals and neurotrophins. The neurotrophins BDNF and NT-3, acting through tyrosine kinase receptors of the Trk family (in particular TrkB and TrkC, often referred to collectively as Ntrk), sustain the survival of spiral ganglion neurons and direct the extension of their peripheral processes towards inner and outer hair cells, with an NT-3 gradient particularly important along the basal-apical axis of the developing cochlea ^[38,39,41]. This system is joined by repulsive and attractive signals of the semaphorin family, including Semaphorin-5B, which intervene in the refinement of peripheral dendritic branches, preventing excessive branching and favouring selective convergence on the appropriate receptor target ^[36,47]. It is worth recalling here the distinction between type I neurons, which innervate almost exclusively the inner hair cells through radial monosynaptic connections, and type II neurons, fewer in number but functionally important in efferent modulation and cochlear nociception, whose axonal trajectory and guidance properties show distinct molecular features still under study ^[41,47].

7.4 Synaptic refinement and the tonotopic map

The final phase of spiral ganglion connectogenesis coincides with functional maturation of the ribbon synapse between the inner hair cell and the type I afferent ending, a process involving a progressive reduction in the number of synaptic contacts per neuron and electrophysiological specialisation related to position along the tonotopic axis of the cochlea ^[42,44]. This refinement is not simply a numerical pruning of contacts but reflects the establishment of a precise spatial correspondence between the peripheral position of the ganglion neuron, the characteristic frequency of the innervated epithelial segment and the central projection to the cochlear nuclei of the brainstem, where frequency topography is preserved and passed on to higher auditory stations ^[42,45]. Single-cell transcriptomic studies of the mouse cochlea have resolved in unprecedented detail the molecular heterogeneity of ganglion neuron subpopulations along the basal-apical axis, providing an interpretive substrate for understanding both the differential vulnerability of these neurons to acoustic insults and the molecular basis of their functional distortion in disease ^[43,40]. Onto this tonotopic framework is grafted, in postnatal life, noise-induced sensorineural hearing loss, in which impaired tonotopic fidelity of the ganglion projections characteristically compromises speech intelligibility in noise, evidence of the functional continuity between the developmental programmes described here and the pathophysiology of the adult ear ^[46,42].

7.5 Clinical implications of ganglion neurogenesis

Detailed knowledge of the stages of delamination, axon guidance and synaptic refinement of the spiral ganglion is not only of descriptive interest: it guides the clinical interpretation of congenital auditory neuropathies and of the cochlear nerve dysgenesis seen on prenatal and postnatal imaging, and provides the biological rationale for experimental neuronal regeneration strategies aimed at restoring connectivity between the cochlear implant and the central nervous system in severe sensorineural hearing loss ^[15,31,29].

8Tonotopy and functional maturation

Tonotopy is the fundamental organising principle of the peripheral auditory apparatus, namely the orderly mapping of characteristic frequencies along the longitudinal axis of the cochlea, from the base, rich in high frequencies, to the apex, tuned to low ones^[42][44][45]. This architecture is not an epiphenomenon of basilar membrane mechanics alone but the result of a programme of molecular regionalisation established early during morphogenesis of the cochlear duct and persisting, as a functional tonotopic map, along the whole central auditory pathway^[14][42].

8.1 Base-to-apex molecular gradients

Along the cochlear axis opposing and complementary signalling gradients are established, essential for the positional specification of hair cells and supporting cells. The retinoic acid gradient, high apically and decreasing towards the base, acts in combination with gradients of bone morphogenetic proteins (BMPs) of opposite polarity, defining a molecular coordinate system that orients the radial and longitudinal differentiation of the sensory epithelium^[16][20][14]. FGF signalling, already crucial for induction of the otic placode and proliferation of prosensory progenitors, contributes further to gradual base-to-apex maturation, modulating the timing of cell cycle exit and the expression of terminal differentiation genes^[9][11][33]. These gradients determine not only the local biomechanical properties of the tectorial and basilar membranes but also the differential expression of ion channels and motor proteins in the outer hair cells, elements that contribute to definitive frequency tuning^[42][43].

8.2 Cochlear morphogenesis and stabilisation of the tonotopic map

Coiling of the cochlear duct, whose definitive three-dimensional architecture emerges from convergent extension and differential growth, is the anatomical substrate onto which the tonotopic map is projected^[14][31][32]. Correct spatial apposition of spiral ganglion neurons to their peripheral targets, guided by axon guidance molecules expressed in gradients, ensures that tonotopic information generated peripherally is preserved in the central projection to the cochlear nucleus^[47][37][39]. The transcription factors governing the specification of different hair cell subtypes, acting in concert with the same morphogenetic gradients, also contribute to a functional regionalisation reflected in the differential distribution of electrophysiological properties along the cochlear axis^[17][15].

8.3 Pre-hearing spontaneous activity

Before the ear canal opens and mechanoelectrical transduction driven by external acoustic stimuli begins, the maturing cochlear epithelium generates intense spontaneous activity, sustained by Kölliker’s organ cells and by calcium currents intrinsic to immature hair cells^[24][34]. This activity, propagated to spiral ganglion neurons through immature glutamatergic synapses, acts as an organising signal that refines connectivity between inner hair cells and type I afferents, in a process closely dependent on the neurotrophic factor neurotrophin-3 and its receptor signalling^[38][41]. In parallel, semaphorin family molecules regulate branching and refinement of the peripheral processes of spiral neurons, helping to define the precision of the neural tonotopic map before the functional onset of hearing^[36][37].

8.4 Synaptic remodelling and the functional onset of hearing

The onset of hearing, which in humans coincides with the end of fetal life, is preceded by a critical phase of synaptic remodelling in which the polyinnervated connectivity characteristic of the immature stage is progressively reduced to a one-to-one relation between inner hair cell and single afferent fibre, a process depending on both intrinsic signals and locally generated spontaneous activity^[24][41][30]. The contribution of programmed cell death at this stage is not negligible: selective elimination of supernumerary neuronal and supporting populations accompanies the definitive refinement of the circuit, as in other phases of labyrinthine morphogenesis^[30][31]. Electrophysiological maturation of the outer hair cells, with acquisition of prestin-mediated electromotility and stabilisation of voltage-dependent potassium currents, marks the transition from generator of spontaneous activity to functional cochlear amplifier, indispensable for the fine frequency selectivity that characterises mature hearing^[42][43][34].

8.5 Clinical relevance of tonotopic distortion

Understanding the mechanisms that stabilise the tonotopic map takes on direct clinical significance in interpreting acquired hearing deficits: acoustic trauma and sensorineural hearing loss can induce reorganisation or distortion of the tonotopic representation both peripherally and centrally, with significant repercussions on speech intelligibility in noise^[46][40]. Single-cell transcriptomic maps of the adult and developing cochlea now offer an unprecedented tool for correlating regional molecular identity with the differential vulnerability of cell populations, opening prospects for targeted therapies based on the specific tonotopic position of the damage^[43][40][45]. In this perspective, thorough knowledge of the morphogenetic gradients that instruct tonotopy in embryonic life is not merely a chapter of developmental biology but a conceptual foundation for pathophysiological interpretation and for developing regenerative strategies aimed at restoring, as far as possible, the original frequency precision of the organ of Corti^[21][22].

9Genetics of congenital deafness

Over the last thirty years our understanding of the molecular mechanisms underlying congenital deafness has been enriched by an ever larger catalogue of genes, whose interpretation requires an integrated reading with the embryology of the inner ear described in the preceding sections^[51,54]. The non-syndromic forms, which account for most prelingual hereditary hearing losses, and the syndromic forms, fewer but of high nosological value, often share the same developmental substrate: differentiation of the cochleovestibular epithelium sensorium from the prosensory domain of the otic placode^[37,17].

9.1 The DFNB1 locus and connexin 26

Mutation of the GJB2 gene, encoding connexin 26, is the most frequent cause of autosomal recessive non-syndromic deafness in European populations, with an effect exerted primarily on the supporting cells and the spiral ligament, structures derived from the otic mesenchyme surrounding the developing organ of Corti^[51,37]. The gap junction defect impairs endolymphatic potassium recycling precisely at the stage when hair cell differentiation, driven by the transcription factor Atoh1, is still proceeding along the basal-apical gradient of the cochlea^[25,27]. The genotype-phenotype correlation at the DFNB1 locus shows how a perturbation apparently confined to the non-sensory compartment can nonetheless result in early functional arrest of the entire sensory epithelium, because of the trophic interdependence between hair cells and supporting cells established during organogenesis^[21,34].

9.2 SLC26A4 and Pendred syndrome

The SLC26A4 gene, which encodes pendrin, is the paradigm of how a single disturbance of endolymphatic ion homeostasis can produce a continuous phenotypic spectrum, from the non-syndromic form DFNB4 to full Pendred syndrome with associated thyroid goitre^[51,54]. Developmentally, pendrin dysfunction is accompanied by enlargement of the vestibular aqueduct and by incomplete cochlear malformations such as the Mondini deformity, direct evidence of interference with the processes of elongation and coiling of the cochlear duct that take place between the fifth and tenth weeks of gestation^[14,32]. This correspondence between genotype and a morphological anomaly documentable on neuroimaging makes SLC26A4 an instructive example of the principle, valid also for other entities such as CHARGE syndrome linked to CHD7, that the radiological cochlear phenotype is a direct indicator of the ontogenetic moment at which the genetic defect manifested itself^[35].

9.3 The molecular motor genes: MYO15A, MYO7A, TMC1

A second major chapter of the genetics of deafness concerns the genes encoding structural and motor proteins of the stereociliary bundle, whose assembly is indispensable for mechanoelectrical transduction^[52,49]. Mutations of MYO15A (DFNB3) arrest stereociliary elongation already at embryonic stages, resulting in a short, disorganised bundle, while variants of MYO7A, besides causing the non-syndromic form DFNB2/DFNA11, are the most common cause of Usher syndrome type 1^[50,53]. The TMC1 gene, an essential component of the mechanoelectrical transduction complex located at the tips of the shorter stereocilia, further illustrates how apparently point-like molecular defects translate into divergent developmental phenotypes depending on whether the mutation is dominant or recessive, with progressive postlingual onset or profound congenital onset respectively^[48,54]. These motor genes share a common denominator with the transcription factors that orchestrate hair cell differentiation during the critical window of cochlear development, underlining the continuity between cell specification and structural maturation of the stereociliary bundle^[17,24].

9.4 Syndromic forms: Usher, Waardenburg, branchio-oto-renal

Usher syndrome, besides MYO7A mutations in type 1, involves genes encoding proteins that anchor and link adjacent stereocilia, whose loss impairs the cohesion of the bundle during its postnatal refinement, with combined consequences for hearing and the retina^[52,53]. Waardenburg syndrome, linked to mutations of PAX3 and its target gene MITF, results from defective migration and survival of the neural crest melanocytes that colonise the stria vascularis, an element crucial for generating the endocochlear potential and hence for the functional rather than morphogenetic maturation of the inner ear^[51,31]. Branchio-oto-renal syndrome, attributable to mutations of EYA1 and its partners in the Six1/Eya1/Pax network, reflects instead an early disturbance of the induction of the otic placode itself and of the associated branchial structures, with malformations ranging from preauricular fistulae to major cochlear dysplasias, consistent with the pleiotropic role of these genes in shared nephrogenesis and otogenesis^[5,8].

9.5 Genotype-development correlation

Taken together, these genetic entities confirm an organising principle: the severity and type of the auditory phenotype depend less on the nosological category of the gene than on the moment and compartment of otic development in which it acts, whether induction of the placode, morphogenesis of the membranous labyrinth, hair cell specification or refinement of the stereociliary bundle^[15,37,54]. This ontogenetic perspective, now enriched by single-cell transcriptional maps of the cochlea, orients molecular diagnosis towards a more accurate prediction of the clinical trajectory and of the useful therapeutic window, including regenerative strategies based on reactivating hair cell developmental programmes^[43,28].

10Clinical correlates

10.1 Labyrinthine malformations: ontogenetic arrest and anatomoradiological classification

Understanding the morphogenetic sequence that leads from the otic placode to the definitive labyrinthine capsule is the interpretive foundation of every inner ear malformation^[6][32]. Michel aplasia is the most radical expression of this pathology, resulting from arrest of otic development around the third week of gestation, before invagination of the otic vesicle, with consequent almost total absence of the cochleovestibular structures^[8][31]. More often the arrest occurs at later stages, when differentiation of the cochlear duct has begun but has not completed its spiral elaboration: this is the case of the Mondini dysplasia, characterised by a cochlea of one and a half turns instead of two and a half, a wide communication between scala vestibuli and scala tympani, and a frequent association with enlargement of the aqueductus vestibuli^[14][31][32]. This enlargement, isolated or associated with other anomalies, reflects persistent patency of the primitive endolymphatic duct and is one of the commonest causes of progressive and fluctuating sensorineural hearing loss in childhood, with a risk of intraoperative perilymphatic gusher^[14][45]. Conceptually distinct is the Scheibe dysplasia, or cochleosaccular dysplasia, in which the ontogenetic arrest is late and selective: the bony labyrinthine capsule develops normally, while the sensory epithelium of the saccule and cochlear duct shows atrophy of the organ of Corti and the stria vascularis, probably through a defect in terminal differentiation of hair cells and supporting cells rather than an anomaly of gross morphogenesis^[24][34][37]. This temporal hierarchy of malformations, aplastic, early dysplastic and late dysplastic, faithfully mirrors the sequence of molecular processes governing otic induction, three-dimensional morphogenesis of the labyrinth and subsequent specification of the sensory epithelia, including the programmed cell death programmes that sculpt the boundaries between compartments^[15][30][31].

10.2 CHARGE syndrome and the CHD7 gene: a model of pleiotropic dysregulation

CHARGE syndrome provides a particularly instructive clinical paradigm of the link between the molecular biology of otic development and complex malformation phenotypes. Mutations of the CHD7 gene, a broad-spectrum chromatin remodeller, alter the transcription of numerous target genes involved in the morphogenesis of the cochlea and vestibule, producing a heterogeneous spectrum ranging from aplasia of the semicircular canals to cochlear hypoplasia with a short cochlea or incomplete turn configuration^[35]. This phenotypic heterogeneity is consistent with the pleiotropic role that signals such as FGF, BMP and Shh play in the spatiotemporal coordination of labyrinthine morphogenesis, and with the sensitivity of these circuits to even partial epigenetic perturbations^[11][18][20]. Comparative analysis of large cohorts of CHARGE patients has made it possible to correlate specific cochlear imaging patterns with different moments of morphogenetic arrest, thus offering a direct bridge between molecular genetics and clinical radiology^[35][1].

10.3 Surgical and functional implications for cochlear implantation

Detailed knowledge of the malformation sequence is not merely descriptive but directly guides cochlear implant planning. In Mondini forms and in enlarged vestibular aqueducts, the abnormal communication between the cerebrospinal and perilymphatic spaces requires a specific surgical strategy for controlling the gusher and positioning the electrode correctly, while variability in the number of cochlear turns affects the choice of array length and the tonotopic estimate after implantation^[14][42][45]. In cases with an intact bony capsule but selective impairment of the sensory epithelium, as in Scheibe dysplasia, survival of the spiral ganglion, dependent during development on neurotrophic signals such as neurotrophin-3, is the main prognostic determinant of implant effectiveness, since it is electrical stimulation of the residual ganglion neurons, not of the absent hair cells, that mediates prosthetic auditory perception^[38][39][41]. Tonotopic organisation too, established during development through molecular gradients along the cochlear axis, affects the correspondence between the physiological frequency map and the stimulation delivered by the electrodes, a crucial element for speech intelligibility in noise in implant users^[42][44][46].

10.4 Gene therapy and hair cell regeneration: translational prospects

The fact that most forms of hereditary sensorineural hearing loss derive from mutations in genes governing the development and function of hair cells, including genes encoding unconventional myosins and structural proteins of the stereociliary bundle, has opened the way to replacement or corrective gene therapy strategies targeted at precise stages of the differentiation cascade^{[49][50][52][54]}. Direct reprogramming with transcription factors such as Atoh1, Gfi1 and Pou4f3, able under embryonic conditions to induce conversion of progenitor cells into hair cells, is today the most promising experimental model for postnatal regeneration, although epigenetic barriers and inhibitory signals mediated by the Notch pathway limit the efficiency of conversion in adult mammals^[23][25][28]. Recent studies on the activation of long-range enhancers needed to maintain the regenerative competence of supporting cells suggest that the optimal therapeutic window may lie at a very early stage of postnatal development, as observed for the plasticity of the Wnt/Notch pathway in the mouse cochlea^[21][22][26]. These findings outline an integrated therapeutic horizon in which early molecular diagnosis, detailed morphological imaging of cochleovestibular malformations and regenerative or replacement strategies converge towards truly personalised management of congenital hearing loss^[17][53].

1General framework: mechanoelectrical transduction as a nanometre-scale event; the hair bundle as an organelle

Of all known sensory processes, mechanoelectrical transduction (mechanoelectrical transduction, MET) has the fastest kinetics and the finest spatial resolution, operating on scales of the order of a nanometre and on times of the order of tens of microseconds ^[3,9]. Understanding the ultrastructure of the cochlear hair cell therefore means accepting a conceptual change of scale: the hair bundle should not be seen as a simple juxtaposition of stereocilia but as a true unitary mechanosensory organelle, whose three-dimensional architecture is itself the functional substrate of the transduction process ^[3,7].

1.1 The hair bundle as a hierarchical system of graded rows

The stereocilia are arranged in rows of increasing height along a strictly encoded graded pattern, connected by filamentous extracellular links that transmit sound-induced force from the tallest stereocilia to the shorter ones ^[1,3]. This organisation is not merely morphological: deflection of the bundle, induced by basilar membrane motion and transmitted through the tectorial membrane, generates mechanical tension along the tip links that translates directly into opening of the transduction channels ^[3,8]. The genetics of the interciliary links, up to five distinct surface specialisations interconnecting stereocilia and kinocilium in the mature bundle of some species, has been the privileged gateway to a functional understanding of this architecture, since mutations in each molecular component produce specific and reproducible ultrastructural phenotypes ^[8,5]. Non-classical cadherins in particular perform functions that go well beyond canonical cell-cell adhesion, acting as essential structural components of the tip link itself and its cytoskeletal anchorage ^[5,8].

1.2 The molecular complex at the site of force transmission

The mechanotransducer channel proper consists of a multiprotein complex whose identity was defined relatively recently: the proteins TMC1 and TMC2, structurally related to the TMEM16 family, have been proposed as the pore-forming subunits of the channel ^[10,11,12]. TMC1/2 alone, however, are not sufficient to reconstitute a functional mechanosensitive channel, and the protein TMIE is indispensable in defining the pore and gating properties of the complex ^[9,12]. This core is joined by LHFPL5, required for maximal channel activation through direct interaction with the tip link, clarifying a mechanical coupling mechanism long poorly defined ^[4,9]. Retention of TMC1 at the precise site of force transmission also depends on LOXHD1, whose absence destabilises the whole transduction apparatus without preventing its initial formation ^[1,9]. An equally critical motor and structural role is played by MYO7A, required both for maintaining the transducing stereocilia and for effective force transmission to the MET channel during cochlear hair cell development; the shaker-1 mouse, carrying a spontaneous missense mutation in the orthologous gene, has for decades been an indispensable model for studying this motor function ^[2,1].

1.3 Intraciliary calcium signalling and its ultrastructural relevance

A less established but increasingly interesting aspect concerns the spontaneous calcium transients observed in the stereocilia of auditory and vestibular hair cells, which suggest local modulation of MET channel opening independent of the primary mechanical stimulus ^[6,9]. This calcium signalling, confined to submicrometre compartments, strengthens the idea that the hair bundle should be regarded as an organelle with internal functional compartmentalisation, not a mere bundle of passive appendages ^[6,3].

1.4 Limits and prospects of in vivo ultrastructural imaging

Experimental verification of these molecular models required the development of imaging techniques able to resolve nanometre structures in living mammalian hair cells, overcoming the limits of conventional light microscopy ^[7,3]. Only through nanoscale-resolution bundle imaging protocols has it been possible to correlate dynamically ciliary deflection, tip link tension and channel complex activation, closing the circle between static ultrastructural data and dynamic electrophysiological phenomena ^[7,4,9]. It is within this methodological framework that the entire conceptual structure of modern molecular otology is set: every non-syndromic hearing deficit of cochlear origin should now be interpreted, in the first instance, as a possible genetic or acquired alteration of one of these discrete ultrastructural elements, each acting on a spatial scale that both challenges and explains the extraordinary sensitivity of the human ear ^[1,9,2].

2The hair bundle and the stereocilia

The hair bundle is the specialised apical apparatus of the cochlear and vestibular hair cell, made up of tens to hundreds of stereocilia arranged in a strictly conserved geometric pattern that governs its entire mechanotransduction physiology^[3,7]. Contrary to historical nomenclature, each stereocilium is not a true cilium with an axoneme of microtubule doublets but an outpouching of the plasma membrane supported by a core of paracrystalline actin filaments, densely packed and stabilised by crosslinking proteins that determine its axial stiffness and anisotropic mechanical behaviour^[3,6].

2.1 The organ-pipe arrangement and the height hierarchy

The most immediately recognisable ultrastructural feature of the bundle is its organisation in rows of increasing height, arranged in a staircase profile or, in the classical description, like organ pipes (staircase pattern), with the tallest row facing the extrastriolar side or, in the cochlea, the lateral wall of the cochlear duct^[3,1]. This hierarchy is not a morphological epiphenomenon but the geometric condition required for the tip links, extracellular filaments connecting the tip of each shorter stereocilium to the side of the adjacent taller one, to transmit the deflecting force along the axis of maximum sensitivity of the bundle^[1,8]. Deflection of the bundle towards the tallest row tensions the tip link and opens the transduction channel, while deflection in the opposite direction relaxes it and closes the channel, according to a gating spring model whose ultrastructural validity has been repeatedly confirmed by nanometre-resolution imaging of living cells^[7,6].

2.2 The actin core and crosslinking proteins

The internal scaffold of the stereocilium consists of 20 to 30 F-actin filaments oriented in parallel with uniform polarity, the barbed end facing the tip, a site of continuous turnover and of anchorage of the transduction complex^[3,2]. The lateral cohesion of this filament bundle is ensured by actin-specific crosslinking proteins, including espin and the espin-like proteins, which form regularly spaced transverse bridges that give the core the torsional stiffness needed to behave as a semirigid rod rather than a compressible bundle^[3,8]. Associated with this scaffold are unconventional molecular motors, in particular myosin VIIa, indispensable not only for the structural maintenance of the shorter transducing stereocilia but also for correct positioning of the mechanotransduction channel complex during development, as shown in shaker-1 mouse models carrying missense mutations in the orthologous gene^[2,1]. Similarly, LOXHD1 has been identified as indispensable for retaining TMC1 channels at the precise point of mechanical force transmission, underlining that the integrity of the actin core cannot be separated functionally from the molecular stability of the transduction machinery it supports^[1,9].

2.3 Basal tapering and the anchoring rootlet

Towards the base the actin core undergoes progressive tapering, with a reduction in the number of filaments, which converge into a thinned structure called the rootlet that penetrates deep into the underlying cuticular plate^[3,7]. This tapered rootlet is the point of maximum concentration of mechanical stress during ciliary deflection and is thought to be the preferential site of fracture under acoustic overstimulation, a finding of direct clinical relevance to the pathogenesis of acoustic trauma^[3,19]. Anchorage of the rootlet to the cuticular plate, itself rich in a dedicated actin network, provides the mechanical constraint that converts distal deflection into a bending moment localised mainly at the base, a behaviour essential for efficient force transmission to the channel complex at the lower end of the tip link^[1,4].

2.4 The transduction complex and accessory tip link proteins

At the lower end of the tip link, on the shorter stereocilium, is concentrated the macromolecular complex responsible for mechanical channel opening, comprising the pore-forming subunits TMC1 and TMC2, functionally and structurally related to the TMEM16 family^[10,11,12], and the regulatory proteins TMIE and LHFPL5, the latter required for maximal channel activation through direct interaction with the gating complex^[9,4]. The tip link itself, of cadherin nature, belongs to the wider family of specialised links that interconnect the stereocilia along their whole length, of which the tip link is only the functionally most critical component for transduction, not the only element of ultrastructural cohesion of the bundle^[5,8].

2.5 The kinocilium: a transient structure and morphogenetic determinant

During development each hair cell has a true kinocilium, the only structure of the bundle with a 9+2 microtubule axoneme, derived from the primary centriole and positioned next to the tallest row of stereocilia^[3,8]. In the mammalian cochlea the kinocilium regresses completely after the perinatal period, but its position during development acts as a determining signal for the planar polarity of the bundle and the orientation of the organ-pipe hierarchy, so that its disappearance leaves a stable imprint on the definitive architecture of the organ of Corti, whereas in the vestibular structures the kinocilium persists in the mature cell, contributing permanently to bundle mechanics^[8,3]. This regional difference in the persistence of the kinocilium underlines how the ultrastructure of the hair bundle, although deriving from a common developmental programme, is remodelled in an organ-specific way to meet distinct functional needs, from high-resolution frequency coding in the cochlea to the detection of linear and angular accelerations in the vestibular systems^[6,2].

3The links of the bundle

3.1 Supramolecular architecture of the stereociliary bundle

The hair bundle of cochlear and vestibular hair cells is not a collection of independent stereocilia but a functional syncytial structure held together by a hierarchy of specialised extracellular links, each with a distinct molecular composition, axial location and biomechanical role^[3,8]. The stereocilia, arranged in rows of increasing height along a gradient strictly encoded during development, are connected not only by the oblique tip link but also by a constellation of lateral junctions that ensure their cohesion during deflection by sound or by linear or angular acceleration^[1,7]. This multi-link architecture allows coordinated distribution of shear forces over the whole bundle, preventing deflection from concentrating harmfully on individual stereocilia and ensuring that the mechanical signal converges synergistically on the transduction channels at the tips of the shorter rows^[3,6].

3.2 The tip link: the CDH23-PCDH15 heterodimer

The tip link is the most studied and functionally crucial filamentous element of the entire mechanotransduction apparatus. It consists of an antiparallel heterotetramer in which two molecules of cadherin-23 (CDH23), arranged as a parallel dimer in the upper segment anchored to the taller stereocilium, interface through a calcium-dependent handshake bond with two molecules of protocadherin-15 (PCDH15), which form the lower segment inserted into the membrane of the adjacent shorter stereocilium, immediately next to the transduction channel complex^[5,8]. This molecular polarity is not random: the lower end of PCDH15 associates directly with the channel components, in particular TMC1/TMC2, LHFPL5 and TMIE, forming a macromolecular complex that converts the mechanical tension generated by bundle deflection into opening of the ion pore^[4,9,12]. The calcium-dependent nature of the CDH23-PCDH15 bond explains the well-known lability of the tip link in the presence of extracellular calcium chelators, a phenomenon historically exploited to dissociate mechanoelectrical transduction selectively in classical electrophysiological studies^[3,5]. Recent nanometre-resolution imaging approaches on living hair cells have made it possible to visualise directly the dynamics of the tip link and its regeneration after mechanical breakage, confirming an active turnover of the cadherin complex compatible with continuous remodelling of the bundle under physiological conditions^[7,6].

3.3 The channel complex and its accessory molecular partners

At the lower end of the tip link, the mechanosensitive pore consists of proteins of the TMC family (TMC1 and TMC2), whose structural homology with the TMEM16 family has suggested a double-pore architecture modulated by membrane lipids^[10,11]. TMC1 and TMC2 do not operate in isolation: the TMIE subunit defines the conductance and gating properties of the pore, while LHFPL5 is required for maximal activation of the complex, acting as a bridging protein between the tip link and the channel itself^[9,4]. The stability of this supramolecular arrangement at the transduction site also depends on LOXHD1, whose absence impairs the localisation of TMC1 in the apical region of the shorter stereocilium, showing that maintenance of the channel and tip link complex requires a dedicated protein scaffold independent of the cadherin bond itself^[1,12].

3.4 Top connectors and lateral links: shaft and ankle links

Above the tip link, in the most apical region of the bundle, lies the top connector, a thin horizontal filament joining the tips of adjacent stereocilia within the same row, contributing to the transverse cohesion of the bundle during deflection and limiting mutual sliding of the stereocilia^[8,3]. Further down the stereociliary axis lie the shaft connectors, which join the shafts at intermediate levels, and at the base, at the interface with the cuticular surface, the ankle links, particularly prominent in the immature stages of bundle development and progressively reduced in the mature hair cell^[8,28]. The molecular composition of these lateral links is dominated by the usherin gene product, in association with VLGR1 (very large G-protein-coupled receptor 1) and whirlin, a PDZ-dependent scaffold protein that stabilises the ankle link complex at the membrane and coordinates its interaction with the underlying actin cytoskeleton^[8,2]. The developmental transition from an immature bundle, richly interconnected by ankle links, to a mature bundle in which the tip link organisation prevails reflects a complex transcriptional maturation programme, as shown by the role of factors such as ZBTB20 in regulating the timing of cochlear epithelial differentiation before the onset of hearing^[28,8].

3.5 The Usher complex as an integrated functional scaffold

Together these molecular interactions form the so-called Usher complex, a protein network that unites tip links, ankle links and transduction machinery into a single functional network whose integrity is a necessary condition for hearing and balance^[8,5]. The molecular motor myosin VIIa (MYO7A), mutated in Usher syndrome type 1B, is essential for maintaining the transducing stereocilia and for force transmission to the MET channel during cochlear development, acting as an active tensioning element of the tip link and channel complex^[2,1]. Mutations in the genes encoding CDH23, PCDH15, usherin, VLGR1 and whirlin are collectively responsible for most cases of Usher syndrome, whose phenotype combines congenital deafness, variable vestibular dysfunction and progressive retinitis pigmentosa, evidence that these same molecular components are shared by the cochlear and vestibular hair bundle and the ciliary structures of retinal photoreception^[8,5,2]. The ultrastructural understanding of these links, made possible by high-resolution dynamic imaging techniques, therefore has direct consequences not only for the physiology of mechanoelectrical transduction but also for the clinical and genetic assessment of syndromic and non-syndromic deafness of stereociliary origin^[7,6,3].

4The mechanotransduction (MET) channel: TMC1/TMC2, TMIE, LHFPL5, CIB2, gating spring and adaptation motor

Auditory mechanotransduction is the central molecular event by which deflection of the hair bundle is converted into a change in ionic conductance at the tips of the shorter stereocilia^[3,7]. The protein complex that achieves this transduction is organised in rows of stereocilia of increasing height, connected by oblique tip links that transmit mechanical force from the point of deflection to the lower end of the adjacent stereocilium, the presumed site of the channel pore^[1,3].

4.1 Molecular architecture of the MET complex

The mechanotransduction complex is not a single channel but a multiprotein assembly in which pore-forming subunits, accessory transmembrane proteins and cytoskeletal components cooperate in close spatial proximity^[8,9]. The non-classical cadherins, in particular protocadherin-15 and cadherin-23, form the extracellular scaffold of the tip link, acting as the primary elastic element that transmits tension to the channel complex^[5,8]. LOXHD1, a PLAT-domain protein associated with the stereociliary membrane, has proved indispensable for retaining TMC1 in situ at the site of force transmission, suggesting a role of structural anchorage rather than that of a simple accessory component^[1].

4.2 TMC1 and TMC2 as pore-forming subunits

The molecular identity of the MET channel pore was long debated, but convergent evidence from electrophysiology, mutagenesis and structural modelling has established TMC1 and TMC2 as the strongest candidates for the subunit forming the conductive pore^[9,10,12]. Structural homology with the TMEM16 family, calcium-activated channels and lipid scramblases, has allowed predictive three-dimensional models of the TMC1 pore conformation to be built, revealing a topology with ten transmembrane segments and a central cavity compatible with the non-selective cation permeation observed experimentally^[11,12]. However, the demonstration that TMC1 and TMC2 alone, in heterologous systems, do not fully recapitulate the biophysical properties of the native channel led to the hypothesis that further regulatory subunits are needed to confer the correct conductance and gating properties^[9,10].

4.3 TMIE and LHFPL5: defining the pore and gating

TMIE (transmembrane inner ear) has been identified as an essential component that defines the pore and gating properties of the channel, acting in concert with TMC1/2 to establish the unitary conductance and voltage dependence of the MET current^[9]. In parallel, LHFPL5 (lipoma HMGIC fusion partner-like 5) is required for maximal channel activation: in its absence the residual MET current shows slowed activation kinetics and reduced mechanical sensitivity, indicating that LHFPL5 acts as an allosteric modulator that effectively couples tip link tension to pore opening^[4,9]. The physical interaction between LHFPL5 and the extracellular domain of TMC1 appears to stabilise a conformation with a higher open probability at rest, a point relevant to understanding the spontaneous activation tone seen in immature hair cells^[4,6].

4.4 CIB2/CIB3 and mechanical coupling to the cytoskeleton

The CIB (calcium and integrin binding) proteins, in particular CIB2, bind to the intracellular N-terminal domain of TMC1, acting as an interface between the channel and the adaptation myosin as well as modulating the channel’s sensitivity to changes in intracellular calcium^[8,9]. This interaction gives the MET complex a dual function: mechanical transduction and dynamic processing of the calcium signal, essential for fast adaptation^[6,8].

4.5 The gating spring and tip link tension

The concept of the gating spring, an elastic element in series with the channel, explains the non-linear relation between bundle displacement and open probability observed in MET current recordings^[3,7]. The tip link, made of protocadherin-15 and cadherin-23 in a heterophilic configuration, is the most plausible molecular correlate of this elastic element, although accessory contributions come from the deformability of the stereociliary membrane itself and from the lateral links (ankle links, top connectors) that redistribute force between adjacent stereocilia^[5,8]. Nanometre-resolution imaging in living cells has made it possible to quantify the submicrometre deformations of the bundle during stimulation, confirming the existence of tension gradients along the axis of the stereociliary rows^[7].

4.6 The adaptation motor: myosin VIIa and myosin 1c

Adaptation of the MET channel, the continuous recalibration of the operating point of the gating spring, depends on a myosin molecular motor anchored to the electron-dense apical plaque of the stereocilium. Myosin VIIa is required not only for the structural maintenance of the transducing stereocilia but also for effective force transmission to the MET channel during cochlear hair cell development, as shown in the shaker-1 mouse model, carrying a spontaneous missense mutation in Myo7a that impairs the motor function of the protein^[2]. Myosin 1c, located at the junction between the lower tip link and the stereociliary membrane, is instead implicated in fast calcium-dependent adaptation, sliding along the actin filaments to release tension after channel opening^[2,6].

4.7 Calcium signalling and spontaneous transients

Calcium permeating the open MET channel plays a dual role: it mediates fast adaptation by binding to CIB2 and to the calmodulin associated with myosin, and it generates spontaneous calcium transients in immature stereocilia, observable even without external mechanical stimulation, a phenomenon reflecting basal channel activity during the critical phases of hair bundle development^[6,4]. Disruption of these mechanisms, whether at the level of the pore (TMC1/TMC2, TMIE), the accessory components (LHFPL5, CIB2, LOXHD1) or the adaptation motor (myosin VIIa), converges clinically on forms of non-syndromic sensorineural deafness and on pictures of auditory neuropathy, underlining the diagnostic importance of fine molecular characterisation of this complex for the clinical genetics of hearing loss^[13,1,9].

5The apical apparatus: cuticular plate, reticular lamina, cell-cell junctions, rootlets in the plate

The apical apparatus of the cochlear and vestibular hair cell is the structural interface between the mechanotransduction domain of the stereociliary bundle and the underlying cell body, and its architecture follows a precise engineering principle: dissipating and redistributing the forces generated by deflection of the bundle without compromising the integrity of the apical pole^[3,7]. The cuticular plate, a dense meshwork of actin crosslinked by crosslinking proteins, anchors the roots of the stereocilia and distributes the mechanical load imparted through the tip links during opening of the transduction channels^[1,3].

5.1 The actin rootlet and anchorage in the cuticular plate

Beneath the apical plasma membrane each stereocilium continues as an axial bundle of actin filaments that penetrates the plate to form the rootlet, a structure whose electron density and degree of compaction directly determine the bending stiffness of the stereocilium at the point of maximum mechanical stress^[3,1]. It is in this region, immediately proximal to the rootlet, that the molecular complexes of transduction are concentrated: LOXHD1 retains the TMC1 channels at the point where force is applied, preventing axial sliding from dissociating the mechanotransduction complex from its cytoskeletal anchorage^[1]. The unconventional myosin MYO7A, an actin-dependent motor located near the plate and along the stereociliary axis, contributes not only to maintaining the resting tension of the tip links but also to the stability of the transducing stereocilia themselves during development, as shown by the phenotype of the shaker-1 mouse^[2]. LHFPL5, an accessory component of the channel complex, acts in this same subrootlet compartment to ensure maximal activation of the mechanosensitive channel, integrating mechanical signals transmitted by the tip link with the geometry of root anchorage^[4]. TMIE, a minor but essential component of the pore, together with these proteins defines the gating properties within the same apical microdomain^[9].

5.2 Molecular identity of the channel and submembrane organisation

The long-debated question of the identity of the mechanotransduction pore has seen TMC1 and its paralogue TMC2 emerge as the strongest candidates for the pore-forming subunits, on the basis of structural homology with the TMEM16 family and convergent functional evidence^[10,11,12]. This molecular architecture cannot be separated from the ultrastructural context of the plate: the channel complex resides in a cytoskeletal microenvironment in which the calcium transients recorded in the stereocilium, correlated with channel opening, require a sequestration and diffusion system compatible with the density of the rootlet and with the proximity of the cuticular plate itself, which acts as a functional reserve of stable actin^[6]. Nanometre-resolution imaging techniques applied to living hair cells have allowed bundle geometry, root insertion and tip link dynamics to be correlated directly, confirming that the apical apparatus is not a static structure but a system dynamically regulated during physiological stimulation^[7].

5.3 The reticular lamina and cell-cell junctions

Above the cuticular plate the apical surface of the hair cell is integrated into the reticular lamina, a mosaic of apical membranes of hair cells and supporting cells sealed by tight and adherens junctions that give the sensory epithelium the tightness needed to separate the endolymphatic and perilymphatic compartments^[19,3]. In this context the non-classical cadherins, protocadherin 15 and cadherin 23, do not only perform the well-known function of forming the tip link but take part more widely in the morphogenesis of the adhesive complexes that stabilise the geometry of the bundle relative to the reticular lamina, in a continuum between classical intercellular adhesion and specialised mechanosensitive adhesion^[5]. The repertoire of extracellular links interconnecting stereocilia and kinocilium, ancillary links, lateral links and kinociliary links, described in detail in genetic reviews of bundle links, helps further stiffen the apical assembly against the tangential stresses imposed by movement of the tectorial membrane^[8].

5.4 Maturation and remodelling of the apical apparatus

The definitive configuration of the apical apparatus is not present ab initio but matures through coordinated remodelling of the cochlear epithelium preceding the functional onset of hearing, a process governed also by transcriptional programmes such as the ZBTB20-dependent one, whose loss impairs the overall structural maturation of the cochlea^[28]. Coordination between retraction of transient links, consolidation of the rootlet in the cuticular plate and settling of the reticular lamina takes place in a critical time window, during which the hair cell progressively assembles its functional elements in an integrated process of functional assembly, from the apical cytoskeleton down to the synaptic base^[19]. Failure to complete this maturation, as seen in models of congenital hypothyroidism, results in a combined disturbance of the maturation of stereociliary links and accessory structures, confirming the close interdependence between the apical apparatus and the overlying extracellular environment^[21].

Overall the apical apparatus is an integrated hierarchical system, rootlet, cuticular plate, cell-cell junctions and reticular lamina, in which each structural level helps preserve the fidelity of mechanical transfer from the stereociliary bundle to the transduction channel, and whose disorganisation, even focal, inevitably results in impaired sensitivity and frequency selectivity of the entire organ of Corti^[3,1,4].

6The ribbon synapse of the inner hair cells

The inner hair cell (IHC) transforms the mechanical signal picked up by the stereociliary bundle into a code of cochlear nerve discharges through a presynaptic specialisation unique among vertebrate chemical synapses: the ribbon synapse. This structure, shared with photoreceptors and retinal bipolar cells, is optimised for sustained, high-frequency and temporally faithful release of glutamate, an indispensable requirement for coding acoustic stimuli that follow one another within microseconds^[14,16]. The ultrastructural understanding of this apparatus has benefited decisively from super-resolution microscopy techniques, in particular STED, which have resolved the spatial arrangement of the ribbon and its relation to the presynaptic density below the diffraction limit^[20].

6.1 Molecular architecture of the presynaptic ribbon

The synaptic ribbon is an electron-dense organelle, spheroidal or elliptical in shape, anchored to the presynaptic plasma membrane at the active zone and surrounded by a halo of synaptic vesicles held in close proximity by protein filaments^[16,17]. Its main structural component is RIBEYE, a protein encoded by the CTBP2 gene through an alternative promoter, which shares a domain homologous to the transcriptional corepressor CtBP2 and a B domain with dehydrogenase-like enzymatic activity capable of self-aggregation, the mechanism underlying assembly of the ribbon itself^[16,19]. The ribbon is anchored to the active zone through interactions with Bassoon, a large structural protein that stabilises its position and governs its release efficiency, as shown by Bassoon-deficient mouse models in which the ribbons detach from the presynaptic membrane and float freely in the cytoplasm^[17,19]. Around the ribbon lie dozens of synaptic vesicles in distinct pools, one tethered to the ribbon itself and one close to the membrane and available for immediate release, whose recruitment dynamics and endocytic recycling determine the sustainability of release during prolonged stimulation^[17,15].

6.2 The CaV1.3 complex and excitation-secretion coupling

Depolarisation of the IHC, generated by the mechanoelectrical transduction current at the apical pole through the TMC1/TMC2 channels and the accessory proteins TMIE and LHFPL5^[9,4], spreads electrotonically to the basal pole, where it activates voltage-gated L-type CaV1.3 calcium channels clustered at the active zone beneath the ribbon^[16,19]. The particular activation kinetics of these channels at relatively hyperpolarised potentials, together with their weak calcium-dependent inactivation, allow a graded calcium influx proportional to the amplitude of the membrane potential, an essential element for analogue transmission of the acoustic signal^[19,16]. The calcium microdomain formed near the channels acts on low-affinity sensors of the vesicles tethered to the ribbon, allowing extremely tight excitation-secretion coupling reproducible cycle after cycle^[16,18]. Metabotropic glutamate receptors, located both pre- and postsynaptically, further modulate this transmission, introducing a level of homeostatic regulation of synaptic excitability similar to that described at retinal ribbon synapses^[18].

6.3 Multivesicular release and temporal coding

A distinctive feature of the cochlear synapse is coordinated multivesicular release, in which several vesicles fuse almost synchronously, generating postsynaptic events of variable amplitude (multiquantal EPSCs) rather than repeated unitary quantal events^[16,15]. This mechanism, whose molecular basis is still debated between compound fusion and vesicle coalescence, ensures high temporal precision of the first spike of the afferent neuron, a necessary condition for binaural sound localisation and for coding the temporal fine structure of the stimulus^[14,13]. The functional diversity of the afferent fibres innervating each IHC, low and high threshold with different spontaneous rates, correlates with a morphological gradient of ribbons along the modiolar-pillar axis of the cell, larger ribbons being associated with higher-threshold fibres^[16,20].

6.4 Auditory synaptopathy and vulnerability to noise

The ribbon synapse is the primary target of noise damage even without a permanent threshold rise, a phenomenon known as hidden cochlear synaptopathy (hidden hearing loss)^[13,15]. Excessive glutamate release induced by intense acoustic stimulation causes excitotoxicity of the afferent postsynaptic terminals, with dendritic swelling and selective loss of ribbon-fibre junctions even before IHC death occurs, a picture clinically related to the forms of auditory neuropathy described in audiological practice^[13,15]. High-threshold, low spontaneous rate fibres, with larger ribbons, are preferentially vulnerable to this insult, a finding with direct implications for understanding the difficulty discriminating speech in noise reported by patients with a normal audiogram^[13,20].

6.5 Ontogenetic maturation of the synapse

The ribbon synapse is not fully functional at birth but matures during a critical perinatal window, during which the number of ribbons per cell falls while their coupling efficiency with CaV1.3 channels rises, in parallel with the onset of hearing^[19,28]. Dedicated transcriptional programmes, such as the one controlled by the factor ZBTB20, orchestrate this synaptic remodelling together with the more general maturation of the cochlear epithelium^[28], while systemic hormonal signals, in particular thyroid hormone, have proved decisive for the correct timing of this process: congenital hypothyroidism in mouse models is associated with delayed maturation of both the tectorial membrane and the ribbon synapses, with functional consequences for transmission of the auditory signal^[21]. This interdependence between structural maturation of the synaptic complex and hormonal status underlines the integrated nature of cochlear development, in which the presynaptic ultrastructure described in this section is the final, not isolated, stage of the chain that transforms the mechanical energy of sound into information coded in the acoustic nerve^[16,21].

7Outer hair cell electromotility: prestin, subsurface cisterns and the cochlear amplifier

In the architecture of the organ of Corti, the outer hair cell (OHC) is the effector of an active amplification mechanism that no other vertebrate sensory epithelium possesses in comparable form. Unlike the inner hair cell, dedicated mainly to afferent transduction and synaptic coding of the acoustic signal^[14,16], the OHC incorporates a somatic motor device that operates in parallel with the mechanoelectrical process of the stereocilia, which nonetheless remains the obligatory first stage of the whole cascade^[3,7].

7.1 Prestin and organisation of the lateral membrane

The molecular substrate of electromotility lies in prestin (SLC26A5), an intrinsic protein of the lateral plasma membrane of the OHC, present at an extraordinarily high density and organised in pseudocrystalline domains whose periodicity was historically inferred from freeze-fracture studies and later reconciled with molecular models of the anion transporter from which prestin evolved^[11,9]. The protein’s conformational sensing, driven by changes in membrane potential and modulated by the intracellular anion concentration, translates into a change in the area of the lateral membrane which, propagated along the cell axis, generates the cyclic shortening and lengthening characteristic of electromotility. Significantly, the efficiency of this mechanism depends on a strictly compartmentalised lipid and protein environment, in many respects comparable, in terms of domain segregation, to the highly specialised organisation governing the anchorage of transduction channels at the stereociliary tip^[1,4].

7.2 The cortical lattice and the spectrin and actin scaffold

Immediately beneath the lateral membrane lies the cortical lattice, a pseudohexagonal network of circumferential actin filaments interconnected by transverse spectrin bridges and anchored to underlying membranous cisterns. This cortical scaffold gives the cell wall the anisotropic stiffness needed for the conformational change of prestin to translate into a coherent axial deformation rather than dissipating in uncontrolled radial deformations, a principle of mechanical transduction in many respects conceptually parallel to that described for the intercellular junctional complexes that constrain the geometry of the stereociliary bundle^[5,8]. The crystallographic regularity of the lattice, its crosslinking density and its degree of coupling with the cortical cytoskeleton vary along the tonotopic axis of the cochlea, probably contributing to the gradation of the axial mechanical properties of the OHC according to characteristic frequency, by analogy with the stiffness gradients documented for other accessory structures of the auditory organ^[26,25].

7.3 The subsurface cisterns

Interposed between the cortical lattice and the lateral plasma membrane, the subsurface cisterns form a system of stacked membranous cisterns in functional, if not always morphological, continuity with the smooth endoplasmic reticulum. Their role is not merely structural: they contribute to calcium compartmentalisation and to local regulation of calcium signalling that indirectly modulates the efficiency of electromechanical coupling, in a context conceptually similar to the spontaneous calcium transients described in the hair bundle itself^[6]. The lamellar arrangement of these cisterns, their regular spacing and their connection to the cortical lattice by pillar filaments form a tripartite system, membrane, cisterns and cytoskeleton, whose mechanical integrity is a necessary though not sufficient condition for effective transmission of the force generated by prestin to the whole cell wall^[11,12].

7.4 The somatic amplifier versus the bundle amplifier

The experimental literature distinguishes two components, not mutually exclusive, of active cochlear amplification: a somatic amplifier, based on the prestin-mediated electromotility just described, and a bundle amplifier, originating in the intrinsic dynamics of the mechanoelectrical transduction complex at the apical pole of the stereociliary bundle, where tip links, TMC1/TMC2 channels, TMIE and the accessory proteins LHFPL5 and TMHS combine to generate active forces able to feed back on the motion of the bundle itself^[9,10,4]. The relative contribution of the two components varies with species, tonotopic position and probably stimulation frequency: at low frequencies the bundle amplifier, operating on time constants compatible with the gating kinetics of the transduction channels, appears dominant, whereas at higher frequencies, where the kinetic limits of gating become restrictive, the somatic amplifier, not bound by the same relation between channel opening and generated force, takes on a predominant role^[12,1]. This operational duality is not purely academic: it affects the interpretation of otoacoustic distortion products and the pathophysiological understanding of deafness due to mutations of genes encoding proteins of the transduction complex^[2,3].

7.5 Mechanical coupling with the tectorial membrane

The mechanical output generated by the somatic amplifier does not end in the outer hair cell but is transmitted, through the tallest stereocilia, to the overlying tectorial membrane, whose mechanical impedance and radial and longitudinal stiffness gradients ultimately determine the frequency tuning of the whole system^[22,23,27]. In this sense cellular electromotility and the viscoelastic properties of the tectorial membrane functionally form a single coupled resonant system, whose conceptual separation into a cellular amplifier and an accessory filter is useful for teaching but incomplete physiologically^[24,25].

8The tectorial membrane: ultrastructure

8.1 General ultrastructural architecture

The tectorial membrane is an acellular, viscoelastic extracellular structure overlying the sensory epithelium of the organ of Corti, and is the distal mechanical load of the cochlear transduction system^[22]. It arises from the interdental cells of the spiral limbus and extends radially to the region of the outer hair cells, with a layered architecture made of type II and IX collagen fibrils organised in parallel bundles embedded in an amorphous glycoprotein matrix^[23]. The radial orientation of the fibrils, shown by electron microscopy and computational biomechanical analysis, has historically been proposed as the anatomical substrate of cochlear frequency analysis through radial shear mechanisms^[24]. Nanometre-resolution imaging of living preparations has made it possible to reconstruct the three-dimensional geometry of the fibrillar structure more faithfully, overcoming the limitations introduced by traditional fixation artefacts^[7].

8.2 Molecular components of the matrix: tectorins, otogelin, CEACAM16

The non-collagenous component of the tectorial matrix is dominated by a family of inner ear-specific glycoproteins, including alpha-tectorin and beta-tectorin, which polymerise to form striated filaments responsible for the structural cohesion of the membrane^[25]. These are joined by otogelin and the adhesion protein CEACAM16, the latter involved in stabilising intermolecular contacts between the fibrillar components and the surrounding gelatinous matrix^[22]. The stoichiometric interaction between these glycoproteins determines the non-linear rheological behaviour of the membrane, with elastic and viscous properties that vary along the tonotopic axis of the cochlea, governing frequency selectivity at each radial section^[26]. Mutations in the genes encoding these matrix proteins produce phenotypes of sensorineural hearing loss with ultrastructurally documentable structural changes, confirming the critical role of molecular composition in the overall mechanical function^[23].

8.3 Hensen’s stripe and Kimura’s membrane

In the innermost part of the tectorial membrane lies Hensen’s stripe, a longitudinal thickening marking the boundary between the limbal region and the region directly overlying the inner hair cells, while in the most lateral region one distinguishes the so-called Kimura’s membrane, corresponding to the marginal portion facing the stereocilia of the outer hair cells^[27]. These regional specialisations are not mere topographic references but reflect local differences in fibrillar density and matrix composition, with direct implications for the differential mechanical coupling of the two hair cell systems^[25]. The radial layering described by Kimura was later confirmed by ultrastructural studies that defined its spatial relations with the tips of the stereocilia of the tallest row of outer hair cells, a crucial point for the transmission of shear forces^[23].

8.4 Stiffness gradients and biomechanical role

Direct measurements of the local stiffness of the tectorial membrane, obtained by microindentation and high-resolution optical imaging, have revealed a longitudinal stiffness gradient decreasing from the base to the apex of the cochlea, parallel to the tonotopic gradient of characteristic frequencies^[26]. This gradient contributes, together with the active properties of the outer hair cells, to local mechanical tuning and cochlear amplification, supporting a model in which the tectorial membrane is not a mere passive load but an active participant in the dynamics of the so-called cochlear sandwich^[27]. Theoretical and experimental models agree in attributing to the tectorial membrane the role of a regional mechanical resonator, able to modulate selectively the transmission of radial shear waves according to the local composition of the matrix^[24].

8.5 Interaction with the stereociliary bundle and mechanoelectrical coupling

Functional contact between the undersurface of the tectorial membrane and the tips of the tallest stereocilia of the outer hair cells is the critical point of transfer of the deflecting forces that open the mechanoelectrical transduction channels^[3]. The tip links, cadherin filamentous connections linking adjacent stereocilia of increasing height, transmit the tension generated by bundle deflection to the mechanosensitive channel complex, thus integrating the extracellular tectorial component with the intracellular transduction apparatus^[5]. Maintaining the integrity of this interface depends on motor proteins such as myosin VIIa and on structural components such as LOXHD1, essential for correct positioning of the TMC1 channels at the site of force transmission^[1][2]. The lateral and apical links of the bundle, classified according to a well-defined genetic taxonomy, complete the picture of surface specialisations regulating tectorial coupling^[8].

8.6 Development and maturation of the tectorial membrane

Structural maturation of the tectorial membrane proceeds in parallel with differentiation of the cochlear epithelium and depends critically on thyroid hormone signalling, whose congenital deficiency produces documented disturbances of fibrillar organisation and delays in ribbon synapse formation^[21]. Specific transcriptional programmes, including the one regulated by the factor ZBTB20, orchestrate the overall maturation of the cochlea, including the extracellular components overlying the sensory epithelium^[28]. Together these processes show that the definitive ultrastructure of the tectorial membrane, with its regional layering and biomechanical gradients, results from the integration of genetic programmes, molecular assembly events and progressive mechanical refinement during postnatal development^[22].

9Otoconia and the otolithic membrane

The otolith organs, the utriculus and sacculus, entrust the transduction of linear accelerations and gravity to a superstructural biomechanical system in which calcium carbonate biocrystals, the otoconia, are embedded in an organised glycoprotein matrix, the otolithic membrane^[29,32]. This composite architecture converts inertial forces into deflection of the stereociliary bundles of the macular hair cells through mechanical coupling that depends critically on the density, geometry and anchorage of the individual crystals^[31,32]. The ultrastructural understanding of this system requires the integration of biochemical data on matrix proteins, morphometric observations of crystal geometry and genetic mouse models that have clarified its mechanisms of biomineralisation.

9.1 Otoconin-90 and the matrix proteins

The main protein component of the otoconial organic matrix is otoconin-90 (OC90), a glycoprotein secreted by the macular supporting cells that has two domains homologous to secretory phospholipase A2, although lacking conserved catalytic enzymatic activity^[30]. This two-domain architecture gives the protein an interaction surface favourable to ion recruitment and templated nucleation of calcium carbonate, placing OC90 in the role of an organising scaffold rather than a simple passive structural component^[29,30]. Matrix recruitment studies have shown that calcium sequestration takes place in a spatially specific way across the macular surface, suggesting a regulated gradient of nucleation that determines the size and topographic distribution of mature otoconia^[29]. Cooperation between OC90 and minor accessory proteins, including components of the otogelin family and cysteine-rich proteins, forms the molecular scaffold on which the calcification process is organised, in conceptual analogy with the biomineralisation mechanisms described in other vertebrate systems^[30,31].

9.2 Otopetrin and ionic regulation of biomineralisation

The gene Otopetrin 1 encodes a membrane channel or transporter essential for otoconia formation, whose inactivation causes complete agenesis of otolithic biomineralisation associated with severe vestibular imbalance in the mouse model^[31]. The otopetrin-null phenotype is accompanied by disturbed organisation of the NADPH oxidase system, suggesting that local redox regulation and proton and ion homeostasis of the supporting cells are indispensable prerequisites for the correct supersaturation ratio of calcium and carbonate ions at the interface between macula and otolithic membrane^[31]. This dependence on the local ionic microenvironment complements conceptually the function of OC90 as an organic template, since calcite nucleation requires not only a suitable protein surface but also chemical conditions of controlled supersaturation, whose disruption in genetic models produces dysplastic otoconia, reduced in number or entirely absent^[29,31].

9.3 Geometry of calcite crystals and organisation of the otoconial mass

Ultrastructurally, mature mammalian otoconia appear as rhombohedral calcite biocrystals organised in a compact but not rigidly crystallised mass resting on the gelatinous surface of the otolithic membrane above the macular epithelium^[32]. Structural analysis of the amphibian vestibular apparatus has described in detail the three-dimensional organisation of the otoconial mass, revealing heterogeneous packing in which the size of individual crystals varies systematically with depth and topographic position relative to the sensory macula^[32]. This size gradient is not random but correlates with the biomechanical properties required for efficient transduction of linear accelerations along different axes, a principle of structured organisation recalling, in functional logic though not in mechanism, the stiffness gradients described in the cochlear tectorial membrane^[26,32]. Phylogenetic comparison offered by the study of benthic elasmobranchs is particularly instructive: in species such as the Port Jackson shark the otoconial organ can incorporate exogenous material together with or in place of endogenous biocrystals, an observation underlining that the functional principle of a dense mass overlying the sensory epithelium is evolutionarily conserved regardless of the chemical origin of the material, endogenous or exogenous, provided the relative density needed for mechanical inertia is ensured^[33].

9.4 Anchoring filaments and continuity with the sensory epithelium

The cohesion of the otoconial mass and its functional adhesion to the macular surface depend on a network of submicrometre filaments that cross the gelatinous layer of the otolithic membrane, anchoring the calcite biocrystals to the underlying columnar layer and ensuring faithful transmission of inertial forces to the stereociliary bundles^[32]. This filamentous organisation is the functional analogue, in the vestibular domain, of the fibrillar connections that in the cochlea link the tectorial membrane to the stereocilia of the outer hair cells, although the molecular composition and geometry of the two systems differ substantially^[22,32]. In both contexts, cochlear and vestibular, the integrity of the overlying extracellular matrix is a necessary condition for the mechanical stimulus, whether acoustic or inertial, to be transmitted without dissipation to the mechanotransduction channels on the shorter stereocilia of the bundle^[1,3].

9.5 Clinical implications of otoconial disorganisation

Partial detachment or degeneration of the otoconial mass, with consequent migration of calcite fragments into the semicircular canals, is the pathophysiological substrate of benign paroxysmal positional vertigo, whose ultrastructural understanding derives directly from the experimental models of otoconial biomineralisation described above^[29,31]. Genetic alterations of the matrix genes or of the otopetrin transporter, although rare as a cause of isolated Mendelian disease in humans, offer in the mouse model an irreplaceable experimental paradigm for dissecting the molecular steps of otolithic calcification and for framing, by pathophysiological extension, the age-related degenerative otoconial mechanisms seen in clinical otoneurological practice^[30,31,33].

10Ultrastructural methods and clinical correlates

Investigation of the cytoarchitecture of the inner ear rests on a layered methodological repertoire whose evolution has progressively shifted attainable resolution from the organelle to the single macromolecule. Characterising the stereociliary bundle, intercellular junctions and ribbon synapses requires integrating classical morphological techniques with cryogenic and computational approaches, since the mechanical fragility of the structures studied, in particular the tip links and transduction complexes, demands extremely controlled fixation and freezing protocols^[3,7].

10.1 Transmission and scanning electron microscopy

Transmission (TEM) and scanning (SEM) electron microscopy remain the foundational tools for describing the arrangement of stereocilia in rows of increasing height and their interconnection by lateral and apical links^[3,8]. High-resolution imaging of living hair cells, using super-resolution approaches applied to the stereociliary bundle, has made it possible to correlate the static morphology obtained after fixation with the dynamics of calcium transients recordable in real time, overcoming the interpretive limits of single-moment images alone^[7,6]. Similarly, the ultrastructure of the ribbon synapse, with its characteristic dense body surrounded by synaptic vesicles, was defined by conventional TEM and later refined by STED microscopy, which revealed the submicrometre organisation of the presynaptic and postsynaptic active zones^[14,20].

10.2 Freeze-fracture and freeze-etch

Rapid freezing and fracture techniques historically provided the only window on the morphology of intercellular links in near-native conditions, avoiding the dehydration artefacts typical of conventional chemical fixation^[8,5]. This approach made it possible to distinguish the five classes of surface specialisation interconnecting stereocilia and kinocilium in the mature bundle, each with distinct biomechanical properties and molecular composition, including the atypical cadherins protocadherin-15 and cadherin-23 as constituents of the tip link filaments^[8,5]. The same cryogenic logic applies to the study of the otoconial matrix, whose biocrystalline organisation and spatial calcium sequestration were clarified using frozen preparations that preserve the interface between matrix protein and mineral phase^[29,30].

10.3 Electron tomography and cryo-EM

The advent of single-particle cryo-electron microscopy and cryo-electron tomography has resolved the three-dimensional architecture of the mechanoelectrical transduction complex at near-atomic resolution, clarifying the structural relation between TMC1 and the proteins of the TMEM16 family^[11,10]. Electrophysiological studies combined with structural modelling have shown that TMC1 and TMC2 are the pore-forming subunits of the mechanosensitive channel, while TMIE defines its pore and gating properties^[12,9]. Tomography has also clarified the role of LHFPL5 in maximal channel activation and the function of LOXHD1 in retaining TMC1 at the site of force transmission, integrating structural data with observations in mutant mouse models^[4,1]. These approaches have definitively resolved the debate, which lasted more than a decade, on the molecular identity of the mechanotransducer channel^[10,11].

10.4 Genotype-phenotype correlations: non-syndromic deafness and Usher syndrome

Integrating ultrastructural data with clinical genetics has made it possible to attribute specific phenotypes of autosomal recessive non-syndromic deafness (DFNB loci) to point defects in constituents of the stereociliary bundle and the transduction complex, as shown by mutations in TMC1^[9,12]. The shaker-1 mouse, carrying a spontaneous missense mutation in Myo7a, has been the paradigmatic model for understanding how the motor deficit of myosin VIIa impairs maintenance of the transducing stereocilia and force transmission to the MET channel during development, correlating directly with Usher syndrome type 1B in humans^[2,3]. The atypical cadherins involved in intercellular links, when mutated, produce overlapping syndromic phenotypes, since their calcium-dependent adhesion function goes beyond the classical intercellular role to take on a mechanotransduction significance specific to the inner ear^[5,8]. In parallel, defects of the ribbon synapse produce the picture of auditory neuropathy, in which synchronous transmission of the acoustic signal is impaired presynaptically despite preserved outer hair cell function^[13,15,16]. Vesicle turnover and presynaptic membrane retrieval, as well as the metabotropic glutamate receptors located at the ribbon synapse, are further molecular targets whose ultrastructural defect has been correlated with clinical forms of sensorineural hearing loss^[17,18,19].

10.5 Hair cell regeneration and therapeutic prospects

The study of cochlear epithelial maturation, including transcriptional remodelling mediated by factors such as ZBTB20, has opened prospects for understanding the molecular programmes needed to acquire auditory competence before functional onset, a finding relevant to regenerative strategies^[28]. The correlation between maturation of the tectorial membrane and of the ribbon synapse in congenital hypothyroidism is a further example of how systemic disturbances can be reflected in multiple concurrent ultrastructural targets^[21,22]. High-resolution dynamic imaging of living hair cells, together with recording of stereociliary calcium transients, now provides the most promising experimental platform for assessing, in regenerative models or cochlear organoids, whether newly formed structures faithfully replicate the molecular architecture of the native bundle^[7,6]. Integrating these ultrastructural methods with genotypic characterisation is therefore the indispensable methodological paradigm for the clinical translation of molecular discoveries in hereditary deafness and in strategies for regenerating the sensory cells of the inner ear.

1General framework: imaging of the temporal bone, the complementary roles of CT and MRI, and indications

Because of the density of functionally critical structures packed into a space of a few millimetres, the temporal bone is one of the most demanding anatomical regions in the whole of neuroradiology. The petrous pyramid houses in succession the external ear, the ossicular chain, the membranous and bony labyrinth, the facial nerve canal and the internal auditory canal, elements whose diagnostic resolution requires imaging techniques that are complementary rather than interchangeable^[1,2,3]. High-resolution computed tomography, now further refined by ultra-high-resolution acquisitions with photon-counting detectors, provides the submillimetre spatial detail needed to assess bone-air and bone-soft tissue interfaces, while magnetic resonance remains irreplaceable for characterising fluid and neural contents, the cochleovestibular nerve and inflammatory or neoplastic processes arising from the cerebellopontine angle^[1,19].

1.1 The principle of complementarity between CT and MRI

The established diagnostic paradigm in temporal bone disease rests on the non-equivalence of the information provided by the two modalities: CT excels in defining bony structures, the walls of the internal auditory canal, the facial canal, the otic capsule and the ossicular chain, whereas MRI is the only technique able to discriminate inflammatory, cholesteatomatous, neoplastic or cerebrospinal fluid tissue with adequate tissue contrast^[2,4]. This dualism should be understood not as redundancy but as integration: in assessing cholesteatoma, for example, CT defines the extent of bone erosion and the surgical landmarks, while diffusion-weighted MRI sequences allow differential diagnosis from post-inflammatory granulation tissue, significantly reducing surgical false negatives^[2]. Similarly, in traumatic disease of the pyramid, isotropic volumetric CT allows multiplanar reconstructions that reveal fracture lines otherwise missed in the pure axial plane, with direct implications for predicting post-traumatic facial palsy and perilymphatic fistula^[5,12].

1.2 Technical evolution: from axial CT to isotropic multiplanar reconstruction

Historically, the spatial resolution of conventional high-resolution CT was an intrinsic limit to full characterisation of the labyrinthine microstructures; the introduction of protocols with submillimetre isotropic voxels and, more recently, photon-counting detectors has redefined reference standards, allowing multiplanar reconstructions and maximum intensity projections without loss of information compared with the primary acquisition plane^[1,11,15]. The first experiences of non-reconstructive multiplanar reformatting date back to the 1980s, when the added value of customised oblique planes in studying the petrous pyramid beyond the classical axial and coronal planes was demonstrated^[14]. This approach was later refined with two-dimensional angled reformatting techniques dedicated specifically to the tympanic and mastoid segments of the facial nerve canal, now an essential component of preoperative planning in otological and neurotological surgery^[12,4]. Standardised geometric alignment of images, now also achievable through automatic segmentation of the lateral semicircular canals, has further improved the inter-individual reproducibility of morphometric measurements, a crucial element for longitudinal comparison and multicentre studies^[9].

1.3 Main clinical indications

Indications for temporal bone imaging span a wide nosological spectrum. In paediatric hearing loss, the combination of high-resolution CT and MRI is the standard for identifying congenital malformations of the bony and membranous labyrinth, from simple cochlear hypoplasia to the most complex forms of labyrinthine dysplasia^[3,6]. In preoperative assessment for cochlear implantation, imaging plays a decisive role both in defining the anatomy of the round window and basal turn of the cochlea and in verifying the integrity and presence of the cochlear nerve, the latter accessible only through high-resolution T2-weighted MRI sequences^[7,8,40]. Superior semicircular canal dehiscence, a clinical and radiological entity increasingly recognised thanks to the spread of ultra-high-resolution CT protocols, requires millimetre characterisation of the arc and length of the dehiscence, with direct prognostic and surgical implications^[10,13]. In Ménière’s disease, MRI with delayed post-contrast sequences has for the first time allowed in vivo visualisation of the morphological substrate of endolymphatic hydrops, turning a historically clinical and audiometric diagnosis into one also supported radiologically^[17,34]. Finally, in lesions of the cerebellopontine angle and internal auditory canal, primarily vestibular schwannoma, contrast-enhanced MRI remains the reference method, with sensitivity and specificity close to absolute, guiding subsequent choices between surveillance, radiosurgery and surgery^[25,27,28].

1.4 Towards an integrated paradigm

The current trend in otological radiology is towards integrated protocols combining, in a single diagnostic session or in close succession, isotropic volumetric CT acquisition and MRI sequences dedicated to the inner ear and cochleovestibular nerve, overcoming the historical dichotomy between bone imaging and soft tissue imaging^[1,19,42]. This integrated approach is particularly relevant in cochlear implant candidacy, where CT alone is insufficient to exclude conditions such as cochlear nerve aplasia or stenosis of the cochlear aperture, entities that substantially change the therapeutic strategy and necessarily require the complementary contribution of MRI^[44,46]. It is in this context of methodological synergy, not mutual substitution, that the whole discussion of the radiological anatomy of the temporal bone in the following chapters should be read.

2High-resolution CT (HRCT) technique: protocols, slice thickness, bone kernel, cone-beam CT, dose and multiplanar reconstructions

2.1 Technical principles and historical limits of spatial resolution

For decades temporal bone imaging suffered from an intrinsic limit: the spatial resolution of conventional high-resolution computed tomography, although better than that of any other skeletal region, remained insufficient for optimal visualisation of submillimetre structures such as the platina stapedis, the tympanic segment of the facial nerve canal and the thin septa of the cellulae mastoideae^[1]. The introduction of ultra-high-resolution scanners, with dedicated detectors and advanced iterative reconstruction, has overcome most of these limitations, offering anatomical definition close to gross histology and extending clinical indications to fine presurgical diagnosis^[1]. In parallel, the technological evolution of cone-beam CT dedicated to the temporal region has made available a low-dose alternative, particularly useful in children and in serial follow-up^[1],[3].

2.2 Acquisition protocols: slice thickness, pitch and bone reconstruction algorithm

The classical protocol for studying the petrous bone involves isotropic volumetric acquisition with a nominal slice thickness between 0.5 and 0.2 mm, a value made possible by the submillimetre-resolution prototypes described in the earliest experiences of postoperative axial imaging^[4]. The choice of a high spatial frequency bone kernel, although it increases image noise, is essential for delineating the cortical margins of the labyrinthus osseus and assessing any thin bony dehiscence, such as that of the superior semicircular canal^[10],[13]. Studies with ultra-high-resolution photon-counting detectors have shown that 0.2 mm slices allow a more accurate estimate of the radiological prevalence of superior semicircular canal dehiscence than traditional protocols, reducing false positives due to partial volume^[10]. Standardisation of scan orientation remains a significant practical problem: variation in patient positioning and in gantry angle requires geometric alignment strategies based on constant anatomical landmarks, such as the lateral semicircular canals, to ensure reproducible comparative measurements between serial examinations or between the two sides^[9].

2.3 Cone-beam CT and dose optimisation

Dedicated temporal cone-beam CT, thanks to its cone-shaped beam and flat detector, allows isotropic acquisitions at a significantly lower effective dose than conventional multidetector CT, while maintaining comparable spatial resolution for bony structures with high intrinsic contrast^[1]. This makes it particularly suitable in children, where assessment of congenital hearing loss often requires repeated examinations over time and where sensitivity to ionising radiation demands careful dosimetric justification^[3]. It should be stressed, however, that the lower tissue contrast of cone-beam CT limits its use in characterising intratympanic soft tissue, an area in which high-resolution multidetector CT retains a diagnostic advantage, particularly in the differential diagnosis of cholesteatoma and its extensions^[2].

2.4 Multiplanar and oblique reconstructions and maximum intensity projections

Isotropic volumetric acquisition has made it possible to move beyond the standard axial and coronal planes in favour of multiplanar reconstructions oriented along the axis of the structures of interest, with a diagnostic gain documented since the first work on non-reconstructive reformatting of the petrous plane^[14]. Two-dimensional angled oblique reconstruction has proved particularly effective for studying the tympanic and mastoid segments of the canalis nervi facialis, allowing the curved course of the nerve to be followed along its whole length in a single plane^[12]. Similarly, multiplanar reconstructions of volumetric data improve depiction of the ossicular chain, the oval window and the round window, with direct implications for surgical planning^[11]. Maximum intensity projections, together with two-dimensional reconstructions, add complementary information in characterising temporal bone lesions, although their interpretation requires specific experience to avoid superimposition artefacts^[15].

2.5 Clinical consequences of the HRCT protocol

The quality of the HRCT protocol directly affects diagnostic accuracy in many clinical scenarios: preoperative assessment for cochlear implantation requires reliable visualisation of the round window and cochlear patency, possible only with adequate slice thickness and bone kernel^{[7],[8],[40],[42],[45]}. Similarly, characterisation of congenital inner ear anomalies in children depends on the protocol’s ability to discriminate micromalformations of the cochlea and vestibule^[6],[3]. In temporal bone trauma, high-resolution multiplanar assessment allows fracture lines to be classified and complications such as ossicular dislocation or involvement of the facial canal to be identified^[5]. Finally, postoperative anatomy, often distorted by tympanoplasty or mastoidectomy cavities, requires a dedicated reconstruction protocol to distinguish expected findings from recurrent disease^[4]. Together these technical elements, submillimetre slice thickness, high-frequency bone kernel, the low-dose cone-beam option and targeted multiplanar reconstructions, form the methodological foundation on which the whole radiological semiology of the temporal bone discussed in the following chapters is based.

3Reference planes: the Pöschl plane and the Stenvers plane

3.1 The intrinsic limitation of conventional orthogonal planes

The labyrinthine anatomy is not organised along the Cartesian axes of the skull but according to the genetically determined oblique orientation of the three semicircular canals relative to the standard axial and coronal planes ^[9]. Thin-slice volumetric acquisitions, now routinely obtained with very high-resolution computed tomography or photon-counting detectors, provide the necessary isotropic dataset, but interpretation on the native axial and coronal planes alone exposes to systematic undersampling of the superior semicircular canal, whose inclination relative to the Frankfort plane varies individually by several degrees ^[1]. The earliest experiences of non-reconstructive multiplanar reformatting of the os petrosum had already shown that dedicated oblique planes allow curved structures to be followed along their real course, overcoming the partial volume artefact that affects rigid orthogonal slices ^[14]. This methodological need found its paradigmatic application in the study of the roof of the superior semicircular canal and, by conceptual symmetry, in the study of the internal acoustic meatus and its long axis.

3.2 The Pöschl plane: geometry and technical rationale

The Pöschl plane is constructed as an oblique reconstruction parallel to the long axis of the superior semicircular canal, obtained by angling the reformatting plane along the course from the anterior ampulla to the common crus, on the basis of an axial reference plane identified manually or by semi-automatic segmentation of the labyrinthine structures ^[9]. Historically the technique derives from the principles of two-dimensional angled reformatting developed for studying the facial nerve canal, where dedicated angulation allowed the tympanic and mastoid segments to be followed along their curved course rather than in fragmentary sections ^[12]. Its conceptual extension to the superior semicircular canal provided, for the first time in a reproducible way, a plane in which the bony roof of the canal appears in its entirety in a single section, an indispensable condition for reliable assessment of its integrity ^[11]. Multiplanar reconstruction of isotropic volumetric data, applied systematically to the petrous block, has made this angulation an integral part of the current diagnostic protocol, clearly distinguishing it from simple review of the raw axial and coronal planes ^[11].

3.3 The Stenvers plane: perpendicular projection and complementary visualisation

The Stenvers plane, historically derived from the conventional posterior oblique radiographic projection, is now reproduced as a reformat perpendicular to the axis of the superior semicircular canal, essentially aligned with the long axis of the petrous pyramid and the course of the internal acoustic meatus ^[14]. Whereas the Pöschl plane shows the canal along its length, the Stenvers plane crosses it in transverse section, thus offering an independent orthogonal check of the bony continuity of the canal roof and reducing the risk of false positives due to oblique beam inclination in the Pöschl plane alone ^[15]. This biplanar complementarity, already valued in the earliest experiences of maximum intensity projection and two-dimensional angled reconstruction applied to temporal bone lesions, remains the methodological foundation for the combined interpretation of the two oblique planes ^[15], ^[4].

3.4 The key clinical application: superior semicircular canal dehiscence

The combined use of the Pöschl and Stenvers planes is now the standard for the radiological diagnosis of superior semicircular canal dehiscence, a condition whose prevalence has recently been redefined thanks to the submillimetre resolution offered by photon-counting tomography ^[10]. Measurement of the arc and length of the dehiscence, parameters correlated with the severity of audiovestibular symptoms in semicircular canal dehiscence syndrome, necessarily requires the longitudinal view offered by the Pöschl plane for reliable quantification ^[13]. Confirmation on the Stenvers plane, orthogonal to the former, substantially reduces the rate of overestimation due to mere bone thinning without true discontinuity, an interpretive error well documented in surgical series correlated with the loud sound stimulation test used in the clinical diagnosis of the syndrome ^[16]. Biplanar integration is therefore not an accessory refinement but an indispensable diagnostic requirement before any therapeutic decision, surgical or conservative, regarding the dehiscence ^[10], ^[13].

3.5 Usefulness in assessing the labyrinthine windows and in implant planning

Although conceived primarily for studying the superior semicircular canal, oblique planes derived from the same geometric principles are also applied in assessing the round and oval windows when planning cochlear implantation, where dedicated angulation improves the preoperative estimate of surgical access to the round window niche ^[8]. Systematic morphometric analysis of the temporal bone on oriented multiplanar reconstructions has shown that the viewing angle significantly affects the estimate of the patency and orientation of the round window relative to the axis of the basal turn ^[40]. This approach forms part of the wider preoperative imaging protocol for cochlear implant candidacy, in which the combination of axial, coronal and dedicated oblique planes remains indicated for complete characterisation of middle and inner ear anatomy ^[42], ^[45], a principle already stated in the first systematic accounts of temporal bone anatomy in relation to implant surgery ^[7].

3.6 Integration with the axial and coronal planes in the reporting protocol

In current reporting practice the Pöschl and Stenvers planes do not replace but complement systematic analysis on the axial and coronal planes, which remain indispensable for assessing the ossicular chain, the tympanic segment of the facial nerve and the oval window in its standard projection ^[4], ^[11]. The availability of ultra-high-resolution isotropic volumetric datasets now allows all these reconstructions to be generated after acquisition without a dedicated rescan, optimising the trade-off between dose delivered and diagnostic information extracted ^[1]. This post-processing flexibility is the natural evolution of the early dedicated angled reformatting techniques, conceived at a time when native multiplanar acquisition was not available and oblique reconstruction had to be planned in advance for each case ^[12], ^[14].

4Systematic CT anatomy of the petrous bone

Systematic analysis of the petrous pyramid by high-resolution computed tomography requires a rigorous compartmental approach, now made possible by ultra-high-resolution protocols with submillimetre isotropic voxels that have progressively replaced conventional acquisitions ^[1]. Multiplanar reconstruction, now indispensable, allows the curved structures of the temporal bone to be followed along dedicated oblique planes, overcoming the intrinsic limits of axial and coronal scanning alone ^[11][14].

4.1 The tympanic cavity and the scutum

The tympanic cavity should be assessed at its three levels, epitympanum, mesotympanum and hypotympanum, with particular attention to the integrity of the scutum, the shield-shaped bony plate bounding the epitympanic recess laterally, whose erosion is the cardinal sign of growing cholesteatomatous disease ^[2]. Assessment of the scutum is integrated with the study of postoperative anatomy, since tympanoplasty and mastoidectomy alter its usual radiological landmarks, requiring systematic comparison with the presurgical anatomy ^[4]. In petrous bone trauma the tympanic cavity should also be inspected for haemotympanum, wall discontinuity and involvement of adjacent structures ^[5].

4.2 The ossicular chain: the incudomalleolar and incudostapedial joints

The ossicular chain should be followed in its anatomical continuity, assessing the relation between the malleus head and the body of the incus at the incudomalleolar joint, typically described as an ice-cream cone image on the axial plane ^[15]. Distally, the incudostapedial joint and the stapes superstructure require reconstructions along the axis of the stapes, often obtained with targeted two-dimensional angulation ^[12]. Ossicular integrity is of primary clinical importance in cochlear implant planning, where the morphometry of the chain and adjacent landmarks determines the surgical approach ^[7][40].

4.3 The epitympanic recess, aditus ad antrum, antrum and mastoid

The anterior and posterior epitympanic recess, the aditus ad antrum and the mastoid antrum form a pneumatic continuum whose patency should be documented systematically, since obstruction of the aditus is an early and sensitive finding of inflammatory or cholesteatomatous disease ^[2][3]. The degree of mastoid pneumatisation, which varies between individuals, should be described in terms of cell extent and relation to the sigmoid sinus and the dura of the middle fossa, elements decisive for planning a safe mastoidectomy ^[4].

4.4 The facial nerve canal: tympanic and mastoid segments

The Fallopian canal should be traced through its labyrinthine, tympanic and mastoid segments, with particular attention to dehiscence of the tympanic segment above the oval window, a frequent and potentially treacherous finding in middle ear surgery ^[12]. The posterior genu and the vertical mastoid segment require dedicated angled reconstructions along the axis of the canal, a technique historically developed precisely to overcome the limits of standard axial and coronal assessment alone ^[12][11]. In cochlear implant surgery the position of the nerve relative to the round window and basal turn determines the surgical approach to cochleostomy ^[8].

4.5 The oval and round windows

The oval window, site of the stapes footplate, and the round window, with its secondary membrane, should be assessed for patency, orientation and relations with the cochlear promontory; visibility of the round window on preoperative CT correlates with ease of surgical access during cochlear implantation ^[8][42]. Congenital anomalies of the oval or round window are frequently associated with other labyrinthine malformations, requiring integrated assessment of the whole inner ear ^[6].

4.6 The vestibule and semicircular canals

The vestibule and the three semicircular canals, superior, posterior and lateral, should be analysed for morphology, calibre and integrity of their bony covering; the lateral semicircular canal is also a standardised landmark useful for geometric alignment of temporal bone CT acquisitions ^[9]. Superior semicircular canal dehiscence, the subject of growing attention with the advent of ultra-high-resolution and photon-counting CT, should be measured in terms of the length and arc of the dehiscence and correlated with the audiovestibular clinical picture ^[10][13]. Functional tests such as loud sound stimulation can complement the morphological finding in diagnosing dehiscence syndrome ^[16].

4.7 The vestibular and cochlear aqueducts

The vestibular aqueduct should be measured at its midpoint according to the classical criteria for pathological enlargement, one of the commonest findings in congenital sensorineural hearing loss ^[6][3]. The cochlear aqueduct, a distinct canal connecting the scala tympani to the subarachnoid space, should be distinguished from the internal auditory canal and assessed for patency, an element relevant to cochlear implant planning and to estimating the risk of perilymphatic gusher ^[45][41]. The calibre of the internal auditory canal itself, and any stenosis, correlates with cochleovestibular nerve deficiency, a crucial aspect in selecting candidates for cochlear or auditory brainstem implantation ^[44][47][19].

4.8 The carotid canal and the jugular bulb

The course of the carotid canal, with its vertical and horizontal segments, and the position of the jugular bulb, often protruding into the hypotympanum or displaced laterally close to the basal turn of the cochlea, are anatomical variants of mandatory surgical relevance in all otological and implant planning ^[40][46]. Systematic documentation of these vascular structures, together with the ossicular, tympanic and labyrinthine assessment described above, completes the standardised preoperative CT report required in modern cochlear implant practice ^[42][43].

5MRI technique: high-resolution T2-weighted sequences (CISS/FIESTA), pre- and post-gadolinium T1, DWI (cholesteatoma), 3D

Magnetic resonance imaging is the indispensable complement to high-resolution computed tomography in studying the ear, since while the latter defines bony architecture in submillimetre detail, MRI shows the membranous content of the labyrinth, the cochleovestibular nerve and the soft tissues of the cerebellopontine angle^[1][19]. The choice of sequence, acquisition plane and contrast agent must answer a precise clinical question, since no single protocol is optimal for all otological indications.

5.1 High-resolution T2-weighted sequences: CISS and FIESTA

The constructive interference in steady state (CISS) and FIESTA sequences are the cornerstone of non-contrast labyrinthine imaging, exploiting the marked fluid sensitivity of the signal to delineate perilymph and endolymph with submillimetre resolution and a high contrast-to-noise ratio^[21][24]. These 3D volumetric sequences allow post-processing multiplanar reconstructions along the axis of the cochlea and semicircular canals, enabling morphometric assessment of the cochlear turns and confirmation of the patency of the internal auditory canal^[19][23]. An intrinsic limitation must, however, be recognised: focal areas of low signal in the vestibule can mimic disease when they actually represent susceptibility artefacts or flow effects, a distinction requiring pathological correlation and technical care^[18][20]. In post-labyrinthitic fibrosis of the posterior semicircular canal, focal loss of fluid signal on these sequences correlates with a documentable functional deficit, extending their diagnostic usefulness beyond mere morphology^[22]. The advent of ultra-high-field scanners has reopened the debate on the real advantage of higher field strength for inner ear imaging, given the marked magnetic inhomogeneity induced by the bone-air interface of the petrous pyramid^[23].

5.2 Pre- and post-gadolinium T1 sequences

T1 sequences with and without gadolinium remain indispensable for characterising space-occupying lesions of the internal auditory canal and cerebellopontine angle, above all vestibular schwannoma, which typically enhances intensely after contrast^[25][26][27]. Combining post-contrast T1 with T2 CISS sequences allows not only diagnosis but also volumetric staging useful for therapeutic planning, whether surgery, radiosurgery or active surveillance^[28][29]. In follow-up after treatment, including surveillance after cochlear implantation in patients previously operated on for schwannoma by the translabyrinthine route, the contrast sequence allows tumour recurrence to be distinguished from scar or granulation tissue^[30]. Appreciable diagnostic quality has also been documented at intermediate and low field strengths, with implications for the accessibility of surveillance imaging in resource-limited settings^[31], while isolated reports of spontaneous tumour regression underline the importance of contrast protocols reproducible over time for reliable longitudinal comparison^[32].

5.3 Diffusion-weighted imaging and cholesteatoma

The introduction of diffusion-weighted imaging (DWI), in particular with non-echo-planar sequences, has revolutionised the non-invasive approach to residual or recurrent cholesteatoma, exploiting the restricted diffusivity of desquamated keratin to generate a characteristic hyperintense signal that distinguishes the cholesteatoma matrix from granulation tissue or chronic inflammation^[2][3]. This application has become particularly important in follow-up after tympanoplasty, reducing the need for purely diagnostic second-look surgery and complementing high-resolution CT imaging of postoperative anatomy^[1][2].

5.4 Imaging of endolymphatic hydrops: delayed post-gadolinium sequences

In assessing Ménière’s disease, delayed acquisition, typically 4 hours after intravenous gadolinium, with heavily weighted 3D T2-FLAIR sequences or sampling perfection with application-optimized contrasts (SPACE) techniques allows direct visualisation of endolymphatic hydrops as an area of non-enhancement within the enhanced perilymphatic space^[17][33][34]. The reproducibility of this method depends critically on the timing of acquisition, the route of administration (intravenous versus intratympanic) and the post-processing protocol used to segment the fluid compartments^[35][38]. Systematic meta-analyses have confirmed a sustained but centre-dependent diagnostic accuracy, attributable mainly to the heterogeneity of acquisition protocols and of semiquantitative scoring of the degree of hydrops^[33][36][37], and recent prospective studies have extended the use of these sequences to the differential diagnosis from vestibular migraine, an area of frequent clinical overlap^[39].

5.5 The integrated protocol in pre-cochlear implant assessment

In cochlear implant candidacy the MRI protocol combines high-resolution 3D T2 sequences for assessing cochlear patency and cochleovestibular nerve integrity with T1 sequences aimed at excluding retrocochlear disease, integrating the morphometric data obtainable from CT^[40][42][45]. In children, where radiation exposure must be minimised, MRI takes on an even more central role, particularly in characterising unilateral deafness and diagnosing cochlear nerve deficiency associated with stenosis of the cochlear aperture^[41][43][44]. It has been repeatedly shown that CT alone, however accurate in bony definition, cannot exclude aplasia or hypoplasia of the cochleovestibular nerve when the internal auditory canal is narrow, making MRI mandatory in this setting before proceeding to implantation, whether cochlear or brainstem^[46][47].

6MRI anatomy of the IAC and CPA

Magnetic resonance is the indispensable complement to high-resolution computed tomography in studying the internal auditory canal (IAC) and the cerebellopontine angle (CPA), since while CT defines the bony labyrinthine architecture and the walls of the canal, only MRI allows direct visualisation of the cochleovestibular nerve, the membranous labyrinth and the cisternal cerebrospinal fluid structures^[19][3]. Heavily T2-weighted sequences at submillimetre resolution, acquired with a 3D technique with long TR and high TE, exploit the natural contrast between hyperintense cerebrospinal fluid and perilymph and hypointense neural and bony structures, allowing detail comparable to gross anatomy^[24][23].

6.1 Anatomy of the internal auditory canal and of the facial and cochleovestibular nerves

The IAC, oriented anteroposteriorly and lateromedially relative to the petrous block, is traditionally divided into the four quadrants described at the lateral part of the meatus, anterosuperior for the facial nerve, posterosuperior for the superior vestibular nerve, anteroinferior for the cochlear nerve and posteroinferior for the inferior vestibular nerve, a compartmentalisation well appreciated on very high-resolution T2 sequences and fundamental in planning auditory brainstem implantation and in the preoperative assessment of narrow canals^[19][47]. In cochlear implant candidates, demonstrating the integrity and continuity of the cochlear nerve within the IAC directly determines the surgical indication, since its aplasia or hypoplasia, sometimes associated with stenosis of the cochlear aperture, correlates with a high prevalence of cochlear nerve deficiency not predictable from bony morphology alone^[44][42][46]. Similarly, in congenitally narrow internal auditory canals, preoperative assessment of the cochleovestibular nerve with very high-resolution sequences predicts with fair accuracy the functional state of the nerve complex, also in view of brainstem implantation^[47][43].

6.2 The cochlea and membranous labyrinth

Volumetric analysis of the membranous labyrinth, made possible by high-resolution 3D fast spin-echo T2 sequences, has shown that cochlear and vestibular dimensions reach their adult configuration as early as the early fetal period, with minimal variability by age, sex and side, a reference point for diagnosing congenital malformations^[24][6]. In children, heavily T2-weighted non-contrast 3D FLAIR sequences allow excellent definition of the membranous labyrinth even in infancy, overcoming the technical difficulties of the small anatomical dimensions^[21][3]. A point of practical relevance concerns areas of signal loss in the vestibule on heavily T2-weighted images, which may correspond either to technical artefacts or to genuine pathological findings, as shown by correlation between histopathology and signal hypointensity, a distinction crucial to avoid mistaking normal variants for hydrops or fibrosis^[18][22]. Fibrosis of the posterior semicircular canal, described as loss of fluid signal associated with a functional deficit, is a paradigmatic example of correlation between MRI finding and histopathological substrate of the membranous labyrinth^[22][20]. Space-occupying lesions of the inner ear, which can mimic endolymphatic hydrops on perilymph-enhanced sequences, must also be distinguished from true hydrops, a confusion that can be avoided only by comparison with heavily T2-weighted sequences^[20][18].

6.3 Endolymphatic hydrops and imaging of Ménière’s disease

The study of endolymphatic hydrops by delayed post-gadolinium MRI, typically 4 hours after intravenous injection, has revolutionised the diagnostic assessment of Ménière’s disease, allowing direct visualisation of distension of the endolymphatic space at the expense of the perilymphatic space^[17][34][33]. 3D-FLAIR sequences and SPACE techniques with variable flip angles allow reproducible differentiation between cochlear and vestibular hydrops, and their systematic use has made it possible to establish semiquantitative grading criteria also used in the differential diagnosis from vestibular migraine^[36][39][38]. The most recent systematic meta-analyses confirm the incremental diagnostic value of delayed acquisition over conventional sequences, while stressing the need to standardise protocols and the delay between injection and acquisition^[37][35].

6.4 The cerebellopontine angle cistern and vestibular schwannoma

The CPA cistern, a subarachnoid space bounded by the pons, the cerebellum and the posterior surface of the petrous bone, contains the cisternal segment of the cochleovestibular nerve between the brainstem and the internal acoustic porus, the preferred site of origin of vestibular schwannoma^[19][25]. MRI with high-resolution T2 sequences and post-contrast T1 acquisitions remains the reference modality for the diagnosis, monitoring and treatment planning of these tumours, with sensitivity and specificity close to one hundred per cent^[27][26][28]. Comparison between intermediate- and high-field scanners, as well as the introduction of very low-field protocols, has shown diagnostically adequate visualisation of schwannoma even at 0.55 T, widening the possibilities of longitudinal surveillance in resource-limited settings^[31][29]. Follow-up of patients under watchful waiting, as well as assessment after translabyrinthine resection with simultaneous cochlear implantation, requires dedicated MRI protocols that take account of artefacts from the implanted device and of the need to exclude residual tumour in the CPA cistern^[30][29]. Isolated reports of spontaneous volumetric regression of vestibular schwannoma further underline the importance of prolonged, standardised MRI monitoring before any invasive therapeutic decision^[32][26].

7Congenital malformations of the ear

Congenital anomalies of the bony and membranous labyrinth are a chapter of primary importance in paediatric diagnostic imaging, since their correct morphological characterisation directly determines eligibility for cochlear implantation and surgical strategy ^[6][3]. The integrated radiological approach, based on combining high-resolution computed tomography of the os temporale with T2-weighted magnetic resonance sequences, makes it possible to distinguish bony malformations from those selectively involving the membranous and neural structures ^[40][42].

7.1 Cochlear dysplasias: from aplasia to Michel dysplasia

The earliest embryogenetic arrest, occurring before the fourth week of otic development, causes complete labyrinthine aplasia, or Michel dysplasia, characterised by total absence of the cochlea, vestibule and semicircular canals, with a petrous bone that is often hypoplastic and without an identifiable internal auditory canal ^[6][3]. Isolated cochlear aplasia, a slightly later form, is distinguished by the persistence of a relatively preserved vestibule and semicircular canals despite total absence of the cochlea, a picture that must be carefully differentiated from a common cavity using submillimetre multiplanar reconstructions ^[1][9]. In these extreme pictures, preoperative assessment must necessarily extend to the search for a residual cochlear nerve, since anatomical absence of the cochlea does not automatically imply absence of the nerve, and vice versa ^[47].

7.2 The common cavity and the spectrum of incomplete partitions

The common cavity, resulting from arrest at about the fifth week of gestation, appears as a single cystic structure without internal differentiation between cochlea and vestibule, well appreciated both on computed tomography and on the heavily T2-weighted sequences used to study the membranous labyrinth in infants ^[21][24]. The spectrum of incomplete partitions represents instead later and subtler arrests: incomplete partition type I shows a cystic cochlea without modiolus or osseous spiral lamina, associated with a dilated vestibule, while incomplete partition type II, corresponding to the classical Mondini malformation, shows a cochlea with a normal basal turn and fusion of the middle and apical turns into a single cavity, typically associated with an enlarged vestibular aqueduct ^[6][3][40]. Incomplete partition type III, linked to mutations of the POU3F4 gene, is characterised by absence of the cribriform plate between the cochlea and the internal auditory canal, with a consequent risk of intraoperative perilymphatic gusher that makes an accurate radiological description before surgery imperative ^[45][41].

7.3 Enlarged vestibular aqueduct

The enlarged vestibular aqueduct, conventionally defined by a diameter greater than 1.5 mm at its midpoint on axial imaging, is one of the commonest bony anomalies found in progressive or fluctuating sensorineural hearing loss in childhood ^[6][3]. Its identification requires dedicated multiplanar reconstructions along the axis of the canal, since standard axial assessment alone may underestimate it ^[11][9]. Its frequent association with the Mondini malformation requires systematic combined assessment of the two anomalies in every protocol for congenital deafness ^[40][42].

7.4 Cochlear nerve hypoplasia and aplasia

Cochlear nerve deficiency, comprising both hypoplasia and aplasia, is a significant cause of severe unilateral sensorineural hearing loss, often associated with a stenotic cochlear aperture of the internal auditory canal ^[44][43]. Its assessment necessarily requires high-resolution magnetic resonance sequences oriented parasagittally to the axis of the canal, since computed tomography alone cannot visualise the neural components directly, a limitation well documented in adult cochlear implant series ^[46][19]. In paediatric cochlear implant candidates, failure to identify a cochlear nerve separate from the other branches of the eighth cranial nerve determines the choice between cochlear and auditory brainstem implantation, as shown in series of auditory brainstem implantation in patients with a narrow internal auditory canal ^[47][41].

7.5 Atresia of the external auditory canal and the Jahrsdoerfer classification

Congenital atresia of the external auditory canal presents with variable degrees of bony hypoplasia, malformation of the ossicular chain and anomalies of the course of the facial nerve, elements that high-resolution computed tomography can document with submillimetre precision ^[1][4]. The scoring system proposed by Jahrsdoerfer, which assigns progressive values to the presence of the oval window, the thickness of the tympanic segment of the facial nerve, mastoid pneumatisation and ossicular morphology, remains the reference tool for predicting the functional benefit of surgical reconstruction ^[42][45]. Structured reporting according to this scheme, integrated with analysis of the course of the facial canal using oblique angled reconstructions along its axis, is now the standard in multidisciplinary planning for the patient with microtia and atresia ^[12][40].

8Acquired disease on imaging

Diagnostic imaging of acquired temporal bone disease requires systematic integration of high-resolution computed tomography and multiparametric magnetic resonance, each modality interrogating distinct anatomical and functional compartments ^[1][11]. While CT provides the spatial resolution needed to characterise bony structures and air-tissue interfaces, MRI provides the tissue contrast indispensable for differentiating inflammatory, neoplastic and labyrinthine fluid tissue ^[2].

8.1 Cholesteatoma and non-EPI diffusion imaging

Middle ear cholesteatoma appears on CT as a soft tissue mass of intermediate density, typically located in the pars flaccida with erosion of the epitympanic scutum, displacement of the ossicular chain and possible erosion of the lateral semicircular canal ^[2][4]. The specificity of CT in the differential diagnosis from other inflammatory masses nonetheless remains limited, justifying the complementary use of MRI ^[2]. The introduction of non-EPI diffusion sequences, in particular the HASTE and PROPELLER/BLADE techniques, represented a qualitative leap compared with traditional EPI sequences, burdened by magnetic susceptibility artefacts in the petrous region ^[2][19]. Cholesteatoma characteristically shows restricted diffusion with marked hyperintensity, allowing distinction from granulation tissue and cholesterol granuloma, a crucial element in surgical planning and in postoperative follow-up for early identification of recurrence without resort to exploratory surgery ^[4][2]. Volumetric multiplanar reconstruction further optimises characterisation of the extent of the lesion into the tympanic recesses and the aditus ad antrum ^[11].

8.2 Otosclerosis: fenestral and retrofenestral foci

Otosclerosis is a primary osteodystrophic process of the otic capsule, divided into fenestral and retrofenestral forms according to the location of the foci of bone remodelling ^[1][11]. The fenestral form, the more common, involves the fissula ante fenestram in front of the oval window, appearing on high-resolution CT as a focal hypodense area that may progress to obliteration of the oval window niche and stapes fixation ^[1]. The retrofenestral form, less common but prognostically more important, involves the labyrinthine capsule around the cochlea, producing the characteristic double ring or pericochlear halo sign, an expression of demineralisation of the otic capsule in its active phase ^[1][9]. Ultra-high-resolution imaging with photon-counting detector technology has substantially improved the ability to detect small foci, previously underestimated with conventional CT protocols, with direct implications for preoperative staging and for predicting the outcome of stapedotomy ^[1][10].

8.3 Vestibular schwannoma: imaging protocol and classification

Vestibular schwannoma is the commonest tumour of the cerebellopontine angle, and MRI with high-resolution T2-weighted sequences combined with post-contrast T1 sequences is the diagnostic gold standard, with sensitivity and specificity close to 100% ^[25][27]. The typical imaging protocol involves high-resolution 3D T2 acquisitions to assess the relation between tumour and cochleovestibular nerve in the internal auditory canal, followed by gadolinium-enhanced T1 sequences to characterise enhancement and exclude submillimetre intracanalicular components ^[26][28]. Size and topographic classification, distinguishing purely intracanalicular forms from those with cisternal extension, guides the choice between imaging surveillance, stereotactic radiosurgery and microsurgical resection, as shown by the randomised V-REX trial comparing wait-and-scan with upfront radiosurgery in small and medium-sized lesions ^[29][25]. Comparison of magnetic field strengths has also shown that even low-field scanners (0.55 T) can provide diagnostically adequate visualisation, widening access to longitudinal follow-up ^[31], while documented cases of spontaneous regression underline the need for standardised surveillance protocols ^[32].

8.4 Endolymphatic hydrops: delayed gadolinium MRI and Nakashima sequences

Direct visualisation of endolymphatic hydrops, the morphological correlate of Ménière’s disease, has become possible thanks to delayed post-gadolinium magnetic resonance techniques developed from the protocols described by Nakashima and colleagues ^[17][34]. The pathophysiological principle rests on the delayed diffusion of gadolinium into the perilymphatic compartment across the blood-labyrinth barrier, with acquisition typically performed 4 hours after intravenous administration of a single or double dose of contrast ^[38][33]. The 3D-FLAIR and 3D-real IR sequences, together with T2-SPACE techniques with different flip angles, allow selective separation of the endolymphatic from the perilymphatic signal, enabling semiquantitative staging of the degree of hydrops in the cochlea and vestibule ^[36][20]. A systematic meta-analysis has confirmed the good diagnostic accuracy of these descriptors, while highlighting methodological heterogeneity between centres in acquisition timing and grading criteria ^[33][37]. The method has also proved useful in the differential diagnosis between Ménière’s disease and functional vestibulopathies such as vestibular migraine, contributing to a more targeted diagnostic pathway in patients with overlapping audiovestibular symptoms ^[39][35]. Careful recognition of signal loss artefacts on heavily T2-weighted sequences is, however, needed, since these can mimic false positive hydrops without real dilatation of the endolymphatic compartment ^[18][20].

9Imaging before and after cochlear implantation

Cochlear implant candidacy is today the clinical setting in which integration between high-resolution computed tomography and magnetic resonance reaches its greatest information density, since the surgical decision depends simultaneously on the bony morphology of the temporal bone, the patency of the membranous labyrinthus and the integrity of the cochleovestibular nerve^[40][42][45]. The standard protocol involves submillimetre volumetric CT of the temporal bone combined with heavily T2-weighted MRI sequences of the inner ear and internal auditory canal, with the aim of stratifying patients according to anatomical criteria that determine both the choice of surgical approach and the length of the electrode to be implanted^[41][46].

9.1 Preoperative morphometric assessment

Morphometric analysis of the temporal bone, including measurement of the basal length of the cochlea, the diameter of the basal turn and the distance between the promontory and the facial nerve, allows personalised planning of electrode insertion depth, reducing the risk of over-insertion or trauma to intracochlear structures^[40][7]. Preoperative assessment of the visibility of the fenestra cochleae on high-resolution CT has taken on a predictive role regarding the surgical ease of access to the round window, with direct implications for the choice between a cochleostomy and a round window approach^[8][7]. In children, the association between congenital inner ear malformations and the risk of perilymphatic gusher or electrode misplacement requires systematic reading of the CT according to the classification schemes of congenital labyrinthine anomalies^[6][3].

9.2 The role of magnetic resonance in candidate selection

Heavily T2-weighted sequences, such as those used for volumetric study of the membranous labyrinth, can identify fibrous or ossifying obliteration of the cochlea, a condition that may escape tomographic assessment alone and that decisively affects surgical strategy^[46][24]. Similarly, assessing the calibre of the internal auditory canal and looking for cochlear nerve deficiency, often associated with stenosis of the cochlear aperture, require combined analysis of axial images and oblique reconstructions parallel to the axis of the eighth cranial nerve^[44][19][47]. In paediatric unilateral deafness, systematic radiological characterisation of associated anomalies is essential for establishing implant eligibility and predicting functional outcome^[43][3].

9.3 Anatomy of the insertion path and relations with the round window

Multiplanar visualisation of the round window recess, obtained with oblique reconstructions along the axis of the basal turn, makes it possible to anticipate the orientation of the electrode relative to the scala tympani and to foresee anatomical obstacles such as a prominent bony crest or an unfavourable angulation of the promontory^[11][7][9]. The availability of angled multiplanar reconstructions has historically improved characterisation of the course of the tympanic and mastoid segments of the facial nerve, an indispensable element in planning the mastoidectomy and posterior tympanotomy that precede electrode placement^[12][4].

9.4 Postoperative imaging: scalar position and verification of insertion

In the postoperative period, high-resolution CT, supplemented where available by ultra-high-resolution techniques, allows the final scalar position of the electrode array to be verified, distinguishing a correct course in the scala tympani from translocation into the scala vestibuli, an event associated with poorer preservation of residual hearing and less favourable functional outcomes^[1][4]. Postoperative axial anatomy of the temporal bone, including assessment of the mastoidectomy cavity, the position of the receiver-stimulator and the intracochlear course of the electrode, is a dedicated chapter of postsurgical radiological semiology^[4][40].

9.5 Complications, surveillance and special situations

Radiological surveillance after cochlear implantation must also deal with complex scenarios, such as coexisting vestibular schwannoma, in which postoperative MRI monitoring requires dedicated protocols for assessing electrical conduction and the morphology of the internal auditory canal in the presence of the device^[30][25]. In patients with a narrow internal auditory canal, intraoperative identification of the cochleovestibular nerve during auditory brainstem implantation has confirmed the predictive value of preoperative imaging in estimating nerve deficiency, reinforcing the indication for combined CT and MRI analysis in all doubtful cases^[47][44]. Finally, the limited sensitivity of computed tomography alone in identifying cochlear nerve anomalies or subtle labyrinthine obliteration justifies maintaining a combined protocol even in high-volume surgical centres, since MRI adds information not redundant with bony assessment alone^[46][42][45].

10Structured reporting and surgical correlates: checklists, radiosurgical landmarks and planning

10.1 Principles of structured reporting

A radiological report of the temporal bone intended for surgical planning cannot be limited to a narrative description: it must become a systematic, reproducible document able to answer the clinical questions posed by the otological and neurotological surgeon¹². Technological progress towards ultra-high-resolution CT and photon-counting protocols has made submillimetre description of middle and inner ear structures possible, but this gain in resolution is clinically exploitable only if accompanied by a reporting format with mandatory items, requiring systematic checking of each relevant anatomical and pathological finding¹⁹. The checklist structure reduces inter-observer variability, standardises communication with the surgeon and minimises the risk of omitting critical findings in information-dense scenarios such as cochlear implant candidacy or staging of vestibular schwannoma⁴⁰⁴².